diff --git a/404.html b/404.html index bc034d95c..d70d2dcd8 100644 --- a/404.html +++ b/404.html @@ -548,6 +548,8 @@

Page not found

Latest

diff --git a/authors/index.xml b/authors/index.xml index aaeff8a66..e19fe5f89 100644 --- a/authors/index.xml +++ b/authors/index.xml @@ -5,7 +5,7 @@ https://oliviergimenez.github.io/authors/ Authors - Source Themes Academic (https://sourcethemes.com/academic/)en-us© Olivier Gimenez 2024Wed, 13 Jul 2022 00:00:00 +0000 + Source Themes Academic (https://sourcethemes.com/academic/)en-us© Olivier Gimenez 2024 https://oliviergimenez.github.io/img/flyfishing.jpg Authors @@ -23,3 +23,9 @@ +ript> + + + + + diff --git a/blog/.DS_Store b/blog/.DS_Store deleted file mode 100644 index 2faf1b743..000000000 Binary files a/blog/.DS_Store and /dev/null differ diff --git a/blog/amanda_sabbatical/index.html b/blog/amanda_sabbatical/index.html index 2b76d724c..5bd955cdf 100644 --- a/blog/amanda_sabbatical/index.html +++ b/blog/amanda_sabbatical/index.html @@ -847,10 +847,10 @@

Related

  • Kick-off meeting of our project on mitigating lynx-vehicle collisions
    • -
  • New paper!
    • -
  • New paper!
    • +
  • New paper!
    • + diff --git a/blog/gdrecostat/featured.gif b/blog/gdrecostat/featured.gif deleted file mode 100644 index 932e3cbf4..000000000 Binary files a/blog/gdrecostat/featured.gif and /dev/null differ diff --git a/blog/ittecop_meeting/index.html b/blog/ittecop_meeting/index.html index 47b88ed3c..be2ea43ea 100644 --- a/blog/ittecop_meeting/index.html +++ b/blog/ittecop_meeting/index.html @@ -839,10 +839,10 @@
    Statistics, ecology & social sciences

    Related

    diff --git a/img/simon.jpg b/img/simon.jpg deleted file mode 100644 index de13aba37..000000000 Binary files a/img/simon.jpg and /dev/null differ diff --git a/img/simon.png b/img/simon.png deleted file mode 100644 index df13fc0e2..000000000 Binary files a/img/simon.png and /dev/null differ diff --git a/index.html b/index.html index 9a3c4cc4e..b22ac0686 100644 --- a/index.html +++ b/index.html @@ -281,7 +281,7 @@ - + diff --git a/index.json b/index.json index c76fc3c11..318c80039 100644 --- a/index.json +++ b/index.json @@ -1 +1 @@ -[{"authors":["admin"],"categories":null,"content":"My name is Olivier Gimenez and I am a CNRS senior scientist at CEFE working at the interface of animal ecology, statistical modeling and social sciences in the Human-Animal Interactions group. I studied maths and statistics in particular, finishing my PhD in 2003. I obtained my Habilitation in ecology and evolution, and more recently embarked in studying sociology.\nWith students and colleagues, we focus on questions in ecology and conservation biology dealing with the impact of climate change and human activities on population dynamics and species distribution and the management of large mammals. I\u0026rsquo;m also interested in human-wildlife interactions and how interdisciplinary approaches may contribute to the coexistence of humans and animals.\nIf you\u0026rsquo;d like to work with us, feel free to contact me, and check out this paper which gives the spirit of how we enjoy doing science. For news and chat, you may find me on Twitter, Mastodon and GitHub 😄. When time allows, I write R posts which are also published on rweekly and r-bloggers.\n","date":-62135596800,"expirydate":-62135596800,"kind":"term","lang":"en","lastmod":-62135596800,"objectID":"2525497d367e79493fd32b198b28f040","permalink":"https://oliviergimenez.github.io/authors/admin/","publishdate":"0001-01-01T00:00:00Z","relpermalink":"/authors/admin/","section":"authors","summary":"My name is Olivier Gimenez and I am a CNRS senior scientist at CEFE working at the interface of animal ecology, statistical modeling and social sciences in the Human-Animal Interactions group.","tags":null,"title":"","type":"authors"},{"authors":null,"categories":null,"content":"\nA list of my former Master students is available here.\nMatthew Silk (post-doc 2022-2023) is a behavioural and disease ecologist researching the links between population dynamics and social structure by developing new social network and demographic models. Matt was a MCSA research fellow. He is now a research fellow in Edinburgh.\nLouis Barbu (intern 2022) was a second year student of BTSA GPN. Louis conducted a study on the functionality of ecological corridors through the European otter in the Hérault department.\nJulie Louvrier (post-doc 2022) is an ecologist with strong quantitative skills. She worked on the interactions between large carnivores using occupancy models.\nMarine Desprez (post-doc 2022) worked on machine learning methods for classification problems in ecology. This was joint work with Vincent Miele.\nAlix Cosquer (post-doc 2019-2022) is an environmental psychologist. She studied how to favor connections between nature and children using educative marine areas, and worked on children\u0026rsquo;s relationships with nature in schools.\nMaud Quéroué (PhD 2018-2021) developed multispecies demographic models to study predator-prey interactions in seabirds and competition/coexistence in passerines. This was joint work with Christophe Barbraud, Pierre-Yves Henry and Frédéric Barraquand. Maud is now a post-doc with Aurélien Besnard.\nValentin Lauret (PhD 2018-2021) developed integrated models for monitoring of bottlenose dolphins in the network of marine protected areas in the French Mediterranean Sea. This was joint work with Hélène Labach from the MIRACETI NGO. Valentin is now a post-doc with Aurélien Besnard.\nHélène Labach (PhD 2017-2021) from GIS3M was working on the conservation of bottlenose dolphins in the North-Western Mediterranean Sea. Her work relied on the data collected through the GDEGeM project. She continues her work as the head of the MIRACETI NGO.\nNina Santostasi (PhD 2016-2019) developed capture-recapture and demographic models for studying hybridization in the wild. As case studies, she considered dolphins in the Gulf of Corinth in Greece (with Giovanni Bearzi and Silvia Bonizzoni from Dolphin Biology and Conservation) and wolves in Italy (with Paolo Ciucci from University of Rome - La Sapienza). Nina is now a post-doc with Paolo Ciucci working on estimating population abundance of brown bears.\nSarah Bauduin (post-doc 2018-2019) built spatially-explicit agent-based models to assess the viability of Eurasian lynx in France, and more broadly to study multispecies demography. Sarah is now a permanent researcher with the French Office of Biodiversity.\nGilles Maurer (PhD 2015-2018, post-doc 2018-2020) studied the interactions between wild and semi-captive pools of Asian elephants with an interdisciplinary framework (demography, economy, anthropology, population genetics). This was joint work with Baptiste Mulot at Beauval Nature, Finn Kjellberg and Nicolas Lescureux from CEFE. Gilles is now a permanent researcher at Beauval Nature, hosted at CEFE.\nLorena Mansilla (PhD 2016-2019) developed capture-recapture and demographic models for social species with wolves, dolphins and elephant seals as case studies. She used spatially-explicit capture-recapture models, social networks, cluster analyses and integrated population models (her thesis was co-supervised with Roger Pradel).\nVincenzo Gervasi (post-doc 2017-2019) used an interdisciplinary approach to understand coexistence patterns and build trans-boundary management tools for coexistence between large carnivores and human activities in Europe. This was joint work with John Linnell from NINA (Norway) and Luigi Boitani from University of Rome - La Sapienza. Vincenzo is now a post-doc with ISPRA back in Italy working on the African swine fever epidemics in wild boar.\nJulie Louvrier (PhD 2015-2018) evaluated and developed occupancy models for assessing the distribution of large carnivores (wolf, lynx) in Europe. This was joint work with Christophe Duchamp, Nolwenn Hoguet-Drouet and Eric Marboutin from ONCFS. Julie is now a post-doc at the Leibniz Institute for Zoo and Wildlife Research.\nLoreleï Guéry (post-doc 2016-2018) studied the dynamics of lynx in Switzerland using spatially-explicit capture-recapture models. This was joint work with Fridolin Zimmermann from KORA. Loreleï then did a post-doc in the MARBEC lab, working with Daniel Gaertner from IRD on the population dynamics of tunas. She is now a permanent researcher at CIRAD.\nBlaise Piédallu (PhD 2013-2016) developed an integrated approach for managing human-wildlife conflicts with the Pyrenean brown bear population as a case study (joint work with Pierre-Yves Quenette from ONCFS). Blaise now works as a biology teacher.\nLaetitia Blanc (PhD 2012-2016, ATER 2016) Worked on the conservation of lynx in France using spatially-explicit individual-based models (joint work with Stephanie Kramer-Schadt from IZW and Eric Marboutin from ONCFS). Laetitia now works as a biology teacher.\nPaméla Lagrange (PhD 2011-2015) studied the drivers of survival and dispersal in Tree Swallows using capture-recapture data (joint work with Marc Bélisle at Sherbrooke). Pam is now working as a project manager at Réserves Naturelles de France.\nTamar Lok (post-doc 2015-2017) developed population models to combine colour-banding, tracking and count data to assess within- and between-individual variation in different aspects of migratory behaviour of the Eurasian Spoonbill and the Red Knot. Tamar is back to the Netherlands where she is a researcher.\nJulien Papaïx (post-doc 2014) assessed the impact of global change on the population dynamics of Atlantic salmons (joint work with Etienne Prévost). Julien is now a permanent researcher at INRA.\nFitsum Abadi (post-doc 2012) developed neat integrated population models. After a post-doc and occupying a lecturer position there, Fitsum is now assistant professor in the US, New Mexico State University.\nSabrina Servanty (post-doc 2008-2009) studied the dynamic of exploited populations with wild boar as a case study. Sabrina then did a post-doc in the US, and a post-doc in St-Pée/Nivelle on the Côte Basque.\nElena Papadatou (post-doc 2009) worked at quantifying environmental variation and its impact on population dynamics of bats. She worked as an environmental expert in Greece for some time, then moved to the UK to take a project manager position at Arup.\nMarlène Gamelon (PhD 2011-2013) studied environmental variation and reproduction tactics in wild boars (joint work with Jean-Michel Gaillard and Eric Baubet). She did a post-doc in Norway. Marlène is now a permanent researcher at CNRS.\nLucile Marescot (PhD 2010-2012, post-doc 2017-2019) analyzed the recolonization of wolves in France (joint work with Eric Marboutin and Guillaume Chapron). She did a post-doc in the US and another post-doc in Germany. Then she developed multi-species occupancy models for large mammals. This was a collaboration with Christophe Duchamp and Nolwenn Hoguet-Drouet from ONCFS (France) and Arnaud Lyet from WWF-US. After working as a freelance data scientist in conservation biology, Lucile got a permanent position as a researcher at CIRAD.\nSarah Cubaynes (PhD 2009-2011, post-doc 2017) developed capture-recapture mixed models (joint work with Christian Lavergne). After a post-doc both in the UK at Oxford and in the US working in the Yellowstone, she did another post-doc with us. Sarah is now assistant professor.\nMathieu Buoro (PhD 2010-2012) worked in evolutionary ecology using mark-recapture data in Atlantic salmons (joint work with Etienne Prévost). Mathieu did a post-doc in the US at Berkeley, then came back as a post-doc in Toulouse. He is now a permanent researcher at INRAE.\nFrançois Guilhaumon (PhD 2008-2010) analysed biodiversity patterns at large scale and worked in conservation biogeography (joint work with David Mouillot). François is now a permanent researcher at IRD.\n","date":1689984000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1689984000,"objectID":"d56f13f3db88f747133e686977ff200c","permalink":"https://oliviergimenez.github.io/people/alumni/","publishdate":"2023-07-22T00:00:00Z","relpermalink":"/people/alumni/","section":"people","summary":"A list of my former Master students is available here.\nMatthew Silk (post-doc 2022-2023) is a behavioural and disease ecologist researching the links between population dynamics and social structure by developing new social network and demographic models.","tags":null,"title":"Former lab members","type":"people"},{"authors":null,"categories":null,"content":" Loutre, Lez, connectivité écologique et médiation scientifique. December 2023. Talk at a meeting between Montpellier city/metropolis and our lab. You can download the slides here and video recording there in French.\nTour d\u0026rsquo;horizon des modèles et de leurs applications. November 2023. Talk a conference co-organised by Cerema, OFB, and centre de ressources TVB et ITTECOP. You can download the slides here in French.\nStatistical ecology, hidden Markov models and the management of large carnivores in Europe. June 28, 2022. Plenary talk for the ISEC2022 conference. You can download the slides and video recording in English.\nDeep learning, false negatives/positives and predator-prey interactions: Peregrinations in R. March 31, 2021. Slides (F = full screen and P = transcript) and video recording in English. All material here. Rencontres R.\nDeep learning, false negatives/positives and predator-prey interactions: Peregrinations in R. March 31, 2021. Slides here (F = full screen and P = transcript) and material there. Video in French. Journées Labex Montpellier.\nLarge carnivores and statistical ecology. Seminar in Lisbon (remotely) Dec 2020. Video and slides.\nFirst steps in Deep Learning - Species identification and camera traps. Speed talk at imaginecology. November 16-17, 2020, Lyon. Video, slides and workflow.\nFlexible spatio-temporal modelling of occupancy using GAMs: A case study with large carnivores. Speed talk at ISEC. 2020, virtual Sydney Australia. Video and slides.\nGrands carnivores et écologie statistique. Séminaire grand public 2020, Rennes France. Vidéo et diapos.\nModélisation spatio-temporelle de l’aire d’occupation : le cas des grands carnivores en France. Annual meeting of the French network in Statistical ecology. 2020, Rennes France. Slides.\nA review of research in capture-recapture: Ecological questions and methods. Wildlife Research and Conservation. 2019, Berlin Germany. Slides, video and material.\nLe projet INTERACT : Interactions entre dauphins et activités humaines sur le littoral méditerranéen français. Rencontres humains - cétacés. 2019, Nice France. Diapos et vidéos.\nInferring animal social networks with imperfect detection. World Marine Mammal Conference. 2019, Barcelona Spain. Slides.\nVariable selection in capture-recapture models using the Lasso. International Statistical Ecology Conference. 2018, St Andrews Scotland. Slides.\nHumains, cétacés et recherche académique. Rencontres sur les Interactions entre Humains et Cétacés. 2018, Marseille France. Diapos et vidéo.\nLe loup dans la presse : approche interdisciplinaire. Séminaire linguistique et écologie. 2018, Montpellier France. Diapos.\nTenter l\u0026rsquo;interdisciplinarité dans le cadre d\u0026rsquo;une thèse sur les conflits humains-ours. Séminaire interdisciplinarité écologie – SHS. 2017, Grenoble France. Diapos et vidéo.\nDealing with many correlated covariates in capture-recapture models. International Statistical Ecology Conference. 2016, Seattle USA. Slides.\nConservation and management of large carnivores in France: A beneficial collaboration between ONCFS practitioners and CNRS researchers. International Conference of Ecological Sciences. 2016, Marseille France. Slides.\nAbundance of bottlenose dolphins along the French Mediterranean coast. International workshop on Bottlenose dolphin conservation and monitoring in the North-Western Mediterranean Sea. 2015, Marseille France. Slides here and there.\nMaking sense of citizen science data: A review of methods. International Congress for Conservation Biology. 2014, Montpellier France. Slides.\nHidden Markov modelling of occupancy data. International Statistical Ecology Conference. 2014, Montpellier France. Slides.\nAll equal, really? Individual variability in capture-recapture models from biological and methodological perspectives. EURING analytical meeting. 2013, Athens USA. Slides.\nEstimating fitness in the wild. International Statistical Ecology Conference. 2012, Oslo Norway. Slides.\nAssessing Inter-population Variation in Survival Using Mark-Recapture Data. Seminar at French National Museum of Natural History. 2011, Paris France. Slides.\nDes modèles de capture-recapture pour l\u0026rsquo;écologie évolutive. Habilitation à Diriger des Recherches. 2010, Montpellier France. Diapos et manuscript.\nCombining longitudinal data and pedigree information to assess heritability of survival. 2009, Pescara Italy. Slides.\nImportance of accounting for detection heterogeneity when estimating abundance: the case of French wolves (Canis lupus). Seminar. 2009, Berkeley USA. Slides.\nIndividual heterogeneity in mortality for incomplete longitudinal data. Workshop in Biodemography. 2009, Stanford USA. Slides.\nPath analysis and structural equation modeling of mark-recapture data. International Statistical Ecology Conference. 2008, St Andrews Scotland. Slides.\nNonparametric estimation of survival using penalized splines: Applications in ecology and evolution. EURING analytical meeting 2007 Dunedin New Zealand, International Biometric Conference 2008 Dublin Ireland, Conference of the French Society for Biometry 2005 Paris France. Slides.\nInvestigating Parameter-Redundancy by Yourself 2003, Radolfzell Germany. Slides.\nEstimation, inference and goodness-of-fit tests for multistate capture-recapture models. PhD defense. 2003, Montpellier France. Slides and manuscript.\n","date":1702252800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1702252800,"objectID":"d3efad569ef55bce750d41b479791de1","permalink":"https://oliviergimenez.github.io/talks/talk/","publishdate":"2023-12-11T00:00:00Z","relpermalink":"/talks/talk/","section":"talks","summary":"Loutre, Lez, connectivité écologique et médiation scientifique. December 2023. Talk at a meeting between Montpellier city/metropolis and our lab. You can download the slides here and video recording there in French.","tags":null,"title":"Talks","type":"talks"},{"authors":null,"categories":null,"content":" Preprints\nMoulherat S., L. Pautrel, G. Debat, M.-P. Etienne, L. Gendron, N. Hautière, J.-P. Tarel, G. Testud, O. Gimenez (2023). Biodiversity Monitoring with Intelligent Sensors: An Integrated Pipeline for Mitigating Animal-Vehicle Collisions. Arpha Preprint.\nBauduin S., Germain E., Zimmermann F., Idelberger S., Herdtfelder M., Heurich M., Kramer-Schadt S., Duchamp C., Drouet-Hoguet N., Morand A., Blanc L., Charbonnel A., Gimenez O. (2022). Eurasian lynx populations in Western Europe: What prospects for the next 50 years? bioRxiv preprint.\nGhislain M., T. Bonnet, U. Godeau, O. Dehorter, O. Gimenez, P.-Y. Henry (2022). Synchrony in adult survival is remarkably strong among common temperate songbirds across France. EcoEvoRxiv preprint.\nLauret V., O. Gimenez, H. Labach, N. Lescureux (2022). Behind the scene of ecological expertise: the place of monitoring for biodiversity workers in the French Mediterranean Marine Protected Areas. SocArXiv preprint.\nViollat L., M. Quéroué, K. Delord, O. Gimenez, C. Barbraud (2023). Bottom-up effects and density dependence drive the dynamic of an antarctic seabird predator-prey system. bioRxiv preprint\nDesprez M., V. Miele, O. Gimenez (2023). Nine tips for ecologists using machine learning. arXiv preprint\nPublished\nMaurer, G., M. Chandelier, B. Mulot, O. Gimenez (2023). Polarized media coverage of conflicting, yet emblematic species: The ambivalent portrayal of the Asian elephant. Biological Conservation 289: 110391. *Co-senior author.\nRigoudy N., A. Benyoub, A. Besnard, C. Birck, Y. Bollet, Y. Bunz, J. Chiffard Carriburu, G. Caussimont, E. Chetouane, P. Cornette, A. Delestrade, N. De Backer, L. Dispan, M. Le Barh, J. Duhayer, J.-F. Elder, J.-B. Fanjul, J. Fonderflick, N. Froustey, M. Garel, W. Gaudry, A. Gérard, O. Gimenez, A. Hemery, A. Hemon, J.-M. Jullien, I. Malafosse, M. Marginean, V. Prunet, J. Rabault, M. Randon, A. Régnier, R. Ribière, J.-C. Ricci, S. Ruette, J. Sentilles, N. Siefert, B. Smith, G. Terpereau, W. Thuiller, A. Uzal, V. Vautrain, G. Dussert, B. Spataro, V. Miele, S. Chamaillé-Jammes (2023). The DeepFaune initiative: a collaborative effort towards the automatic identification of French fauna in camera-trap images. European Journal of Wildlife Research 69: 113.\nBauduin S., O. Gimenez, N. Drouet-Hoguet, J. Louvrier, C. Duchamp (2023). Estimer l’effort d’échantillonnage de réseaux participatifs : l’exemple du réseau Loup-lynx. Biodiversité 5: 6-9.\nNakamura M., J. V. López-Bao, H. Rio-Maior, S. Roque, P. Gil, A. Serronha, E. García, O. Hernández Palacios, G. Ferrão da Costa, F. Álvares, F. Petrucci-Fonseca, O. Gimenez*, P. Monterroso* (2023). Insights into the dynamics of wolf occupancy in human-dominated landscapes. Biological Conservation 286: 110316. *Co-senior author.\nForcada J., Hoffman J. I., Gimenez O., Staniland I. J., Bucktrout P., \u0026amp; Wood A. G. (2023). Ninety years of change, from commercial extinction to recovery, range expansion and decline for Antarctic fur seals at South Georgia. Global Change Biology 00, 1–21.\nLauret V., H. Labach, L. David, M. Authier, O. Gimenez (2023). Using integrated multispecies occupancy models to map co-occurrence between bottlenose dolphins and fisheries in the Gulf of Lion, French Mediterranean Sea. Oikos e10270.\nCouturier T., S. Bauduin, G. Astruc, A. Blanck, C. Canonne, T. Chambert, J. Chiffard, A. Cosquer, S. Cubaynes, L. Curtet, E. Dortel, N. Drouet-Hoguet, C. Duchamp, C. Francesiaz, O. Grente, A. Jailloux, M. Kervellec, V. Lauret, J.-D. Lebreton, J. Louvrier, L. Marescot, R. Mathevet, M.-L. Navas, C. Perrot, N. Poulet, P.-Y. Quenette, M. Salas, G. Souchay, C. Vanpé, A. Besnard, O. Gimenez (2023). Building spaces of interactions between researchers and managers: Case studies with wildlife monitoring and conservation in France. Ecological Solutions and Evidence 4:e12245.\nSilk M., O. Gimenez (2023). Generation and applications of simulated datasets to integrate social network and demographic analyses. Ecology and Evolution 13:e9871.\nKervellec M., C. Milleret, C. Vanpé, P.-Y. Quenette, J. Sentilles, S. Palazón, I. A. Jordana, R. Jato, M. M. Elósegui Irurtia, O. Gimenez (2023). Integrating opportunistic and structured non-invasive surveys with spatial capture-recapture models to map connectivity of the Pyrenean brown bear population. Biological Conservation 278: 109875.\nFernández-López, J., P. Acevedo, O. Gimenez (2023). La unión hace la fuerza: modelos de distribución de especies integrando diferentes fuentes de datos. Ecosistemas 32: 2527-2527.\nRodrigues P., I. Dorresteijn, O. Gimenez (2023). ‘The human shield effect’: Human-wildlife co-occurrence patterns in the coffee forests of southwestern Ethiopia. Food Webs 36: e00288.\nPetridou M., J.F. Benson, O. Gimenez, Y. Iliopoulos, V. Kati (2023). Do husbandry practices reduce depredation of free-ranging livestock? A case study with wolves in Greece. Biological Conservation 283: 11009.\nKleiven E.F., F. Barraquand, Gimenez O., J.-A. Henden, R. Anker Ims, E. M. Soininen, N. G. Yoccoz (2023). A Dynamic Occupancy Model for Interacting Species with Two Spatial Scales. JABES In press.\nGrente O., C. Duchamp, S. Bauduin, S. Chamaille-Jammes, N. Drouet-Hoguet, O. Gimenez (2023). Tirs dérogatoires de loups en France : évaluation des effets sur les dommages aux troupeaux. Paru dans Naturae numéro 5.\nTardy C., D. Ody, O. Gimenez, S. Planes (2023). Abundance of fin whales (Balaenoptera physalus) in the north-western Mediterranean Sea, using photo-identification and microsatellite genotyping. Marine Ecology 44:e12737.\nPetridou M., J.F. Benson, O. Gimenez, V. Kati (2023). Spatiotemporal Patterns of Wolves, and Sympatric Predators and Prey Relative to Human Disturbance in Northwestern Greece. Diversity 15:184.\nFrost F., R. McCrea, R. King, O. Gimenez, E. Zipkin (2023). Integrated Population Models: Achieving Their Potential. Journal of Statistical Theory and Practice 17:6.\nLauret V., H. Labach, D. Turek, S. Laran, O. Gimenez (2022). Integrated spatial models foster complementarity between monitoring programmes in producing large‐scale bottlenose dolphin indicators. Animal Conservation 26: 228-238.\nVanpé C., B. Piédallu, P.-Y. Quenette, J. Sentilles, G. Queney, S. Palazón, I. A. Jordana, R. Jato, M. M. Elósegui Irurtia, J. Solà de la Torre, O. Gimenez (2022). Estimating abundance of a recovering transboundary brown bear population with capture-recapture models. Peer Community Journal 2:e71.\nAdamantopoulou S., A. A. Karamanlidis, P. Dendrinos, O. Gimenez (2022). Citizen science indicates significant range recovery and defines new conservation priorities for Earth\u0026rsquo;s most endangered pinniped in Greece. Animal Conservation 26: 115-125.\nNaciri, M, J Aars, MA Blanchet, O. Gimenez, S Cubaynes (2022). Reproductive senescence in polar bears in a variable environment Frontiers in Ecology and Evolution.\nGimenez O., M. Kervellec, J.-B. Fanjul, A. Chaine, L. Marescot, Y. Bollet, C. Duchamp (2021) Trade-off between deep learning for species identification and inference about predator-prey co-occurrence: Reproducible R workflow integrating models in computer vision and ecological statistics. Computo. https://doi.org/10.57750/yfm2-5f45. Data and code here.\nGrente O., T. Saubusse, Gimenez O., E. Marboutin, C. Duchamp (2022). Wolf depredation hotspots in France: Clustering analyses accounting for livestock availability. Biological Conservation 267: 109495.\nGervasi V., J. Linnell, T. Berce, L. Boitani, B. Cretois, P. Ciucci, C. Duchamp, A. Gastineau, O. Grente, D. Hilfiker, D. Huber, Y. Iliopoulos, A. Karamanlidis, F. Marucco, Y. Mertzanis, P. Mannil, H. Norberg, N. Pagon, L. Pedrotti, P.-Y. Quenette, S. Reljic, V. Salvatori, T. Talvi, M. von Arx, Gimenez O. (2021). Ecological correlates of large carnivore depredation on sheep in Europe. Global Ecology and Conservation 30: e01798.\nKazakou E., Fried G., Cheptou P.-O., and Gimenez O. (2021). Does seed mass drive interspecies variation in the effect of management practices on weed demography? Ecology and Evolution 00: 1– 9.\nMaurer G., Gimenez O., Mulot B., and Lescureux N. (2021). Under pressure: How human-wild-captive elephant social-ecological system in Laos is teetering due to global forces and sociocultural changes. People and Nature 3: 1047-1063.\nRounsevell M.D.A., A. Arneth, C. Brown, W.W.L. Cheung, O. Gimenez, I. Holman, P. Leadley, C. Luján, S. Mahevas, I. Maréchaux, R. Pélissier, P. H. Verburg, G. Vieilledent, B.A. Wintle, Y-J Shin (2021). Identifying uncertainties in scenarios and models of socio-ecological systems in support of decision-making. One Earth 4: 967-985.\nTavecchia G., O. Gimenez, R. Choquet, A. Rotger, D. Oro, S. Tenan, A. Sanz-Aguilar (2021). The trap of hidden processes: Why ‘quick \u0026amp; dirty’methods to estimate mortality are not always good. A comment to De Pascalis et al.(2020). Biological Conservation 256: 109057.\nSantostasi N. L., O. Gimenez, R. Caniglia, E. Fabbri, L. Molinari, W. Reggioni, P. Ciucci (2021). Estimating Admixture at the Population Scale: Taking Imperfect Detectability and Uncertainty in Hybrid Classification Seriously. Journal of Wildlife Management 85: 1031-1046.\nZriki G., R. Blatrix, D. J. Bicout, O. Gimenez, A.‐S. Soulié, L. Dadu, D. Degueldre, G. Chiron, N. Sleeckx, L. Roy (2021). Population‐level impact of native arthropod predators on the poultry red mite Dermanyssus gallinae. Journal of Experimental Zoology Part A: Ecological and Integrative Physiology 335: 552-563.\nMiele V., S. Dray, O. Gimenez (2021) Images, écologie et deep learning.. Regard 95 de la Société Française d\u0026rsquo;Ecologie et d\u0026rsquo;Evolution.\nTurek, D., C. Wehrhahn, Gimenez O. (2021). Bayesian Non-Parametric Detection Heterogeneity in Ecological Models. Environmental and Ecological Statistics 28: 355-381.\nLauret V., Labach H., Authier M., Gimenez O. (2021). Combining single- and repeated-visit occupancy models to make the best of monitoring surveys. Ecology 102: e03535.\nGamelon M, Baubet E., Besnard A., Gaillard J.-M., Lebreton J.-D., Touzot L., Veylit L., Gimenez O. (2021). Efficient use of harvest data: An integrated population model for exploited animal populations. Ecography 44: 1296-1310.\nLabach H., Azzinari C., Barbier M., Cesarini C., Daniel B., David L., Dhermain F., Di-Méglio N., Guichard B., Jourdan J., Robert N., Roul M., Tomasi N., Gimenez O. (2022). Distribution and abundance of bottlenose dolphin (Tursiops truncatus) over the French Mediterranean continental shelf. Marine Mammal Science 38: 212-222.\nBarraquand F., Gimenez O. (2021). Fitting stochastic predator-prey models using both population density and kill rate data Theoretical Population Biology 138: 1-27.\nCubaynes S., Aars J., Yoccoz N.G., Pradel R., Wiig Ø, Ims R.A., Gimenez O. (2021). Modeling the demography of species providing extended parental care: A capture-recapture approach with a case study on Polar Bears (Ursus maritimus). Ecology and Evolution 11: 3380-3392.\nQuéroué M., C. Barbraud, F. Barraquand, D. Turek, K. Delord, N. Pacoureau, Gimenez O. (2021). Multispecies integrated population model reveals bottom-up dynamics in a seabird predator-prey system Ecological Monographs 91: e01459.\nGrente O., C. Duchamp, S. Bauduin, T. Opitz, S. Chamaillé-Jammes, N. Drouet-Hoguet, O. Gimenez (2020) Tirs dérogatoires de loups en France: état des connaissances et des enjeux pour la gestion des attaques aux troupeaux. Paru dans Faune Sauvage numéro 327.\nBischof R., C. Milleret, P. Dupont, J. Chipperfield, M. Tourani, A. Ordiz, P. de Valpine, D. Turek, J. A. Royle, Gimenez O., Ø. Flagstad, M. Åkesson, L. Svensson, H. Brøseth, J. Kindberg (2020). Estimating and forecasting spatial population dynamics of apex predators using transnational genetic monitoring. Proceedings of the National Academy of Sciences 117:30531-30538.\nVitet C., P. Duncan, Gimenez O., C. Mabika, S. Chamaillé‐Jammes (2020). Plains zebras bring evidence that dilution and detection effects may not always matter behaviorally and demographically. Ecosphere 11: e03288.\nBauduin S., O. Grente, N.L. Santostasi, P. Ciucci, C. Duchamp, Gimenez O. (2020). An individual-based model to explore the impacts of lesser-known social dynamics on wolf populations. Ecological Modelling 433: 109209.\nMcClintock B.T., R. Langrock, Gimenez O., E. Cam, D.L. Borchers, R. Glennie, T. A. Patterson (2020). Uncovering ecological state dynamics with hidden Markov models. Ecology Letters 23: 1878-1903.\nGenovart M., Gimenez O., A. Bertolero, R. Choquet, D. Oro, R. Pradel (2020). Decrease in social cohesion in a colonial seabird under a perturbation regime. Scientific reports 10: 18720.\nSantostasi N.L., Ciucci P., Bearzi G., Bonizzoni S., Gimenez O. (2020). Assessing the dynamics of hybridization through a matrix modelling approach. Ecological Modelling 431: 109120.\nChero G., Pradel R., Derville S., Bonneville C., Gimenez O., Garrigue C. (2020). Reproductive capacity of an endangered and recovering population of humpback whales in the Southern Hemisphere. Marine Ecology Progress Series 643: 219-227.\nMilleret C., Dupont P., Chipperfield J., Turek D., Brøseth H., Gimenez O., de Valpine P., Bischof R. (2020). Estimating abundance with interruptions in data collection using open population spatial capture–recapture models. Ecosphere 11( 6):e03172.\nLouvrier L., Papaïx J., Duchamp C., Gimenez O. (2020). A mechanistic-statistical species distribution model to explain and forecast wolf (Canis lupus) colonization in South-Eastern France. Spatial Statistics 36: 100428.\nMarescot L., Lyet A., Singh R., Carter N., Gimenez O. (2019). Inferring wildlife poaching in southeast Asia with multispecies dynamic occupancy models. Ecography 42: 1–12.\nGamble A., Garnier R., Chambert T., Gimenez O., Boulinier T. (2020). Next generation serology: integrating cross-sectional and capture-recapture approaches to infer disease dynamics. Ecology 101:e02923.\nBarraquand F., Gimenez O. (2019). Integrating multiple data sources to fit matrix population models for interacting species. Ecological Modelling 411: 108713.\nGimenez O., Gatti S., Duchamp C., Germain E., Laurent A., Zimmermann F., Marboutin E. (2019). Spatial density estimates of Eurasian lynx (Lynx lynx) in the French Jura and Vosges Mountains. Ecology and Evolution 9: 11707-11715.\nRenner, I.W., Louvrier J., Gimenez O. (2019). Combining multiple data sources in species distribution models while accounting for spatial dependence and overfitting with combined penalized likelihood maximization. Methods in Ecology and Evolution 10: 2118-2128.\nBonnet‐Lebrun A. , Karamanlidis A. A., de Gabriel Hernando M. , Renner I. and O. Gimenez (2019). Identifying priority conservation areas for a recovering brown bear population in Greece using citizen science data. Animal Conservation 23: 83-93\nDonfrancesco V., Ciucci P., [14 authors], Gimenez O., [28 authors] and Mukherjee N. (2019). Unravelling the Scientific Debate on How to Address Wolf-Dog Hybridization in Europe. Frontiers in Ecology and Evolution 7:175. doi: 10.3389/fevo.2019.00175.\nGervasi V., Linnell J.D.C., Brøseth H., Gimenez O. (2019). Failure to coordinate management in transboundary populations hinders the achievement of national management goals: The case of wolverines in Scandinavia. Journal of Applied Ecology 56: 1905-1915.\nLok T., Hassell C.J., Piersma T., Pradel R. Gimenez O. (2019). Accounting for heterogeneity when estimating stopover duration, timing and population size of red knots along the Luannan Coast of Bohai Bay, China. Ecology and Evolution 9: 6176–6188.\nLe Bot T., A. Lescroël, J. Fort, C. Péron, O. Gimenez, P. Provost, D. Grémillet (2019). Fishery discards do not compensate natural prey shortage in Northern gannets from the English Channel. Biological Conservation 236: 375-384.\nDorinda Marie Folio, J. Aars, O. Gimenez, A. E. Derocher, Ø. Wiig and S. Cubaynes (2019). How many cubs can a mum nurse? Maternal age and size influence litter size in polar bears. Biology Letters. 15: 20190070.\nDupont P., Milleret C, Gimenez O., Bischof R (2019). Population closure and the bias‐precision trade‐off in spatial capture–recapture. Methods in Ecology and Evolution 10: 661-672.\nSantostasi N.L., P. Ciucci, R. Caniglia, E. Fabbri, L. Molinari, W. Reggioni, Gimenez O. (2019). Use of hidden Markov capture–recapture models to estimate abundance in the presence of uncertainty: Application to the estimation of prevalence of hybrids in animal populations. * Ecology and evolution * 9: 744-755.\nGimenez O., L. Mansilla, M.J. Klaich, M.A. Coscarella, S.N. Pedraza, E.A. Crespo (2019). Inferring animal social networks with imperfect detection. Ecological Modelling. 401: 69-74.\nMelovski, D., Von Arx, M., Avukatov, V., Breitenmoser-Würsten, C., Đurović, M., Elezi, R., Gimenez O., \u0026hellip; , Breitenmoser, U. (2019). Using questionnaire surveys and occupancy modelling to identify conservation priorities for the Critically Endangered Balkan lynx Lynx lynx balcanicus. Oryx. 54: 706-714.\nBenhaiem S., Marescot L., East M. L., Kramer-Schadt S., Gimenez O., Lebreton J.-D. Hofer H. (2018). Slow recovery from a disease epidemic in the spotted hyena, a keystone social carnivore Communications Biology. 201: 1.\nLouvrier J., Chambert T., Marboutin E., Gimenez O. (2018). Accounting for misidentification and heterogeneity in occupancy studies using hidden Markov models. Ecological Modelling. 387: 61-69.\nSantostasi N. L., Bonizzoni S., Gimenez O., Eddy L., Bearzi G. (2018). Common dolphins in the Gulf of Corinth are Critically Endangered. Aquatic Conserv: Mar Freshw Ecosys. 31: 101-109. https://doi.org/10.1002/aqc.2963.\nLouvrier J., Molinari‐Jobin A., Kéry M., Chambert T., Miller D., Zimmermann F., Marboutin E., Molinari P., Mueller O., Cerne R., Gimenez O. (2019). Use of ambiguous detections to improve estimates from species distribution models. Conservation Biology. 33: 185-195.\nBenhaiem S., Marescot L., Hofer H., East M.L., Lebreton J.-D., Kramer-Schadt S., Gimenez O. (2018). Robustness of eco-epidemiological capture-recapture parameter estimates to variation in infection state uncertainty. Frontiers in Veterinary Science 5: 197.\nGimenez O., Lebreton J.‐D., Choquet R., Pradel R. (2018). R2ucare: An r package to perform goodness‐of‐fit tests for capture–recapture models. Methods in Ecology and Evolution 9: 1749–1754.\nSalguero‐Gomez R., Violle C., Gimenez O., Childs D. (2018). Delivering the promises of trait‐based approaches to the needs of demographic approaches, and vice versa. Functional Ecology 32: 1424–1435.\nBancila R. I., Pradel R., Choquet R., Plaiasu R., Gimenez O. (2018). Using temporary emigration to inform movement behaviour of cave‐dwelling invertebrates: a case study of a cave harvestman species. Ecological Entomology 43: 551-559.\nChandelier M., Steuckardt A., Mathevet R., Diwersy S., Gimenez O. (2018). Content analysis of newspaper coverage of wolf recolonization in France using structural topic modeling Biological Conservationa 220: 254-261.\nMarescot L, Benhaiem S, Gimenez O. et al. (2018). Social status mediates the fitness costs of infection with canine distemper virus in Serengeti spotted hyenas. Functional Ecology 32: 1237–1250.\nBarbraud C., Bertrand A., Bouchón M., Chaigneau A., Delord K., Demarcq H., Gimenez O., Torero M. G., Gutiérrez D., Oliveros-Ramos R., Passuni G., Tremblay Y., Bertrand S. (2018). Density dependence, prey accessibility and prey depletion by fisheries drive Peruvian seabird population dynamics. Ecography 41: 1092-1102.\nMarzolin G., Gimenez O. (2018). Fitting a Gamma-Gompertz survival model to capture-recapture data collected on free-ranging animal populations. Journal of Open Source Software. 3(21): 216.\nBearzi G., Gimenez O. (2018). Searching for meaning in marine mammal shared data. Ethics in Science and Environmental Politics. 18: 9-13.\nHamel S., Gaillard J.-M., Yoccoz N. G., Bassar R. D., Bouwhuis S., Caswell H., Douhard M., Gangloff E. J., Gimenez O., Lee P. C., Smallegange I. M., Steiner U. K., Vedde, O. and Vindenes Y. (2018). General conclusion to the special issue Moving forward on individual heterogeneity. Oikos. 127: 750–756.\nDesprez M., Gimenez O., McMahon C.R., Hindell M.A., Harcourt R.G. (2018). Optimizing lifetime reproductive output: Intermittent breeding as a tactic for females in a long-lived, multiparous mammal. Journal of Animal Ecology. 87: 199–211.\nBumb I., Garnier E., Coq S., Nahmani J., Del Rey Granado M., Gimenez O., E. Kazakou (2018). Traits determining the digestibility– decomposability relationships in species from Mediterranean rangelands. Annals of Botany. 121: 459–469.\nRobardet E., Borel C., Moinet M., Jouan D., Wasniewski M., Barrat J., Boue F., Monchatre-Leroy E., Servat A., Gimenez O., Cliquet F., Picard-Meyer E. (2017). Long-term population surveys of two serotine bat (Eptesicus serotinus) colonies exposed to EBLV-1 (European Bat Lyssavirus type 1): Assessment of rabies transmission using capture-recapture models. PLoS Neglected Tropical Diseases. 11 (11), e0006048.\nGimenez O., E. Cam, J.-M. Gaillard (2018). Individual heterogeneity and capture–recapture models: what, why and how? Oikos. 127: 664–686.\nGimenez O., J.-M. Gaillard (2018). Estimating individual fitness in the wild using capture–recapture data. Population Ecology. 60: 101–109.\nGimenez O., C. Barbraud (2017). Dealing with many correlated covariates in capture–recapture models. Population Ecology. 59: 287-291.\nBertulli C.G., L. Guéry, N. McGinty, A. Suzuki, N. Brannan, T. Marques, M. H. Rasmussen, Gimenez O. (2017). Capture-recapture abundance and survival estimates of three cetacean species in Icelandic coastal waters using trained scientist-volunteers. Journal of Sea Research. 131: 22-31.\nMaurer G., Rashford B. S., Chanthavong V., Mulot B., Gimenez O. (2017). Wild-captive interactions and economics drive dynamics of Asian elephants in Laos. Scientific Reports. DOI:10.1038/s41598-017-13907-x.\nLieury N., Devillard S., Besnard A., Gimenez O., Hameau O., Ponchon C., Millon A. (2017). Designing cost-effective capture-recapture surveys for improving the monitoring of survival in bird populations. Biological Conservation. 214: 233-241.\nTidière M., Lemaître J.-F., Pélabon C., Gimenez O. and Gaillard, J.-M. (2017). Evolutionary allometry reveals a shift in selection pressure on male horn size. Journal of Evolutionary Biology. 30: 1826–1835.\nLagrange P., Gimenez O., Doligez B., Pradel R., Garant D., Pelletier F., Bélisle M. (2017). Assessment of individual and conspecific reproductive success as determinants of breeding dispersal of female tree swallows: A capture-recapture approach. Ecology and Evolution. 7: 7334-7346.\nLouvrier, J., C. Duchamp, V. Lauret, E. Marboutin, S. Cubaynes, R. Choquet, C. Miquel, O. Gimenez (2018). Mapping and explaining wolf recolonization in France using dynamic occupancy models and opportunistic data. Ecography. 41: 647-660.\nPiédallu, B., P.-Y. Quenette, N. Bombillon, A. Gastineau, C. Miquel, O. Gimenez (2019). Determinants and patterns of habitat use by the brown bear Ursus arctos in the French Pyrenees revealed by occupancy modelling. Oryx. 53: 334–343.\nSantostasi, N.L., S. Bonizzoni, G. Bearzi, L. Eddy, O. Gimenez (2016). A Robust Design Capture-Recapture Analysis of Abundance, Survival and Temporary Emigration of Three Odontocete Species in the Gulf of Corinth, Greece. Plos One. 11(12): e0166650.\nKazakou, E., G. Fried, J. Richarte, O. Gimenez, C. Violle and A. Metay (2016). A plant trait-based response-and-effect framework to assess vineyard inter-row soil management. Botany Letters. 163: 373-388.\nPéron, G., J.-M. Gaillard, C. Barbraud, C. Bonenfant, A. Charmantier, R. Choquet, T. Coulson, V. Grosbois, A. Loison, G. Marzolin, N. Owen-Smith, D. Pardo, F. Plard, R. Pradel, C. Toïgo, O. Gimenez (2016). Evidence of reduced individual heterogeneity in adult survival of long-lived species. Evolution. 70: 2909-2914.\nAbadi, F., C. Barbraud, O. Gimene (2017). Integrated population modeling reveals the impact of climate on survival of juvenile emperor penguin. Global Change Biology. 23: 1353-1359.\nMcCrea, R.S., B.J.T. Morgan, O. Gimenez (2017). A new strategy for diagnostic model assessment in capture-recapture. Journal of the Royal Statistical Society: Series C. 66: 815-831.\nTidière, M., J.-M. Gaillard, V. Berger, D.W.H. Müller, L. Bingaman Lackey, O. Gimenez, M. Clauss and J.-F. Lemaître (2016). Comparative analyses of longevity and senescence reveal variable survival benefits of living in zoos across mammals. Scientific Reports. 6: 36361.\nBearzi, G., S. Bonizzoni, N.L. Santostasi, N.B. Furey, L. Eddy, V.D. Valavanis, O. Gimenez (2016). Dolphins in a Scaled-Down Mediterranean: The Gulf of Corinth\u0026rsquo;s Odontocetes. Advances in Marine Biology. 75: 297-331.\nSoldatini, C., Y.V. Albores-Barajas, B. Massa and O. Gimenez (2016). Forecasting ocean warming impacts on seabird demography: a case study on the European storm petrel. Marine Ecology Progress Series. 552: 255–269.\nGervasi, V., Brøseth, H, Gimenez, O., Nilsen, E.B., Odden, J., Flagstad, O. and J.D.C. Linnell (2016). Sharing data improves monitoring of trans-boundary populations: the case of wolverines in central Scandinavia. Wildlife Biology. 22: 95–106.\nPiédallu, B., Quenette, P.-Y., Mounet, C., Lescureux, N., Borelli-Massines, M., Dubarry, E., Camarra, J.-J. \u0026amp; Gimenez, O. (2016). Spatial variation in public attitudes towards brown bears in the French Pyrénées. Biological Conservation. 197: 90–97.\nBoulenger, C., A. Acou, O. Gimenez, F. Charrier, J. Tremblay \u0026amp; E. Feunteun (2016). Factors underlying survival of yellow European eels in two unexploited eel sub-populations. Freshwater Biology. 61: 947-962.\nGager, Y., O. Gimenez, M. Teage O\u0026rsquo;Mara and Dina K.N. Dechmann (2016) Group size, survival and surprisingly short lifespan in socially foraging bats. BMC Ecology. 16:2.\nC.A. Bost, C. Cotté, P. Terray, C. Barbraud, C. Bon, K. Delord, O. Gimenez, Y. Handrich, Y. Naito, C. Guinet, H. Weimerskirch (2015). Large-scale climatic anomalies affect marine predator foraging behaviour and demography. Nature Communications. 6: 8220.\nTidière, M., J.-M. Gaillard, D.W.H. Müller, L. Bingaman Lackey, O. Gimenez, M. Clauss, J.-F. Lemaître (2015). Does sexual selection shape sex-differences in longevity and senescence patterns across vertebrates? A review and new insights from captive ruminants. Evolution. 69: 3123–3140 .\nMouquet, N., Lagadeuc, Y., Devictor, V., Doyen, L., Duputié, A., Eveillard, D., Faure, D., Garnier, E., Gimenez, O., Huneman, P., Jabot, F., Jarne, P., Joly, D., Julliard, R., Kéfi, S., Kergoat, G. J., Lavorel, S., Le Gall, L., Meslin, L., Morand, S., Morin, X., Morlon, H., Pinay, G., Pradel, R., Schurr, F. M., Thuiller, W., Loreau, M. (2015). Predictive ecology in a changing world. Journal of Applied Ecology. 52: 1293–1310.\nGamelon, M., Gaillard, J.-M., Gimenez, O., Coulson, T., Tuljapurkar, S., Baubet, E. (2016). Linking demographic responses and life history tactics from longitudinal data in mammals. *Oikos. *125: 395–404.\nKaramanlidis, A.A., M. de Gabriel Hernando, L. Krambokoukis, O. Gimenez (2015). Evidence of a large carnivore population recovery: counting bears in Greece. Journal for Nature Conservation. 27: 10-17.\nBischof, R., H. Broseth, O. Gimenez (2016). Wildlife in a politically divided world: insularism inflates estimates of brown bear abundance. Conservation Letters. 9: 122–130.\nGimenez, O., Buckland, S.T., Morgan, B.J.T., Bez, N., Bertrand, S., Choquet, R., Dray, S., Etienne, M.-P., Fewster, R., Gosselin, F., Mérigot, B., Monestiez, P., Morales, J., Mortier, F., Munoz, F., Ovaskainen, O., Pavoine, S., Pradel, R., Schurr, F.M., Thomas, L., Thuiller, W., Trenkel, V., de Valpine P., E. Rexstad (2015). Statistical ecology comes of age. Biology Letters. 10: 20140698. Data and R codes.\nTidière, M., J.-M. Gaillard, D.W.H. Müller, L. Bingaman Lackey, O. Gimenez, M. Clauss, J.-F. Lemaître (2014). Males do not senesce faster in large herbivores with highly seasonal rut. Experimental Gerontology. 60: 167–172.\nKoons, D.N., F. Colchero, K. Hersey, and O. Gimenez (2015). Disentangling the effects of climate, density dependence, and harvest on an iconic large herbivore\u0026rsquo;s population dynamics. Ecological Applications 25: 956–967.\nGervasi, V., Brøseth, H., Gimenez, O., Nilsen, E., Linnell, J. (2014). The risks of learning: confounding detection and demographic trend when using count-based indices for population monitoring. Ecology and Evolution. 4: 4637–4648.\nGamelon, M., Gimenez, O., Baubet, E., Coulson, T., Tuljapurkar, S., Gaillard, J.-M. (2014). Influence of Life-History Tactics on Transient Dynamics: A Comparative Analysis across Mammalian Populations. The American Naturalist 184: 673-683.\nBarnagaud, J.Y., Papaïx, J., Gimenez, O., Svenning, J.-C. (2014). Dynamic spatial interactions between the native invader Brown-headed Cowbird and its hosts - a metacommunity approach. Diversity and Distributions. 21: 511-522.\nOrgeret, F, Garrigue, C., Gimenez, O., Pradel, R. (2014). Robust assessment of population trends in marine mammals applied to the New Caledonia Humpback Whales. Marine Ecology Progress Series 515: 265–273.\nKoons, D.N., M. Gamelon, J.-M. Gaillard, L.M. Aubry, R.F. Rockwell, F. Klein, R. Choquet and O. Gimenez (2014). Methods for Studying Cause-Specific Senescence in the Wild. Methods in Ecology and Evolution 5: 924-933.\nGamelon, M., S. Focardi, J.-M. Gaillard, O. Gimenez, C. Bonenfant, B. Franzetti, R. Choquet, F. Ronchi, E. Baubet, J.-F. Lemaître. (2014). Do age-specific survival patterns of wild boar fit current evolutionary theories of senescence? Evolution. 68: 3636–3643.\nBlanc, L., E. Marboutin, S. Gatti, F. Zimmermann and O. Gimenez (2014). Improving abundance estimation by combining capture-recapture and occupancy data: example with a large carnivore. Journal of Applied Ecology. 51: 1733–1739.\nGimenez, O., L. Blanc, A. Besnard, R. Pradel, P. F. Doherty Jr, E. Marboutin and R. Choquet (2014). Fitting occupancy models with E-SURGE: hidden-Markov modelling of presence-absence data. Methods in Ecology and Evolution. 5: 592–597. Wiki.\nLee, K.A., C. Huveneers, O. Gimenez, V. Peddemors, R.G. Harcourt (2014). To catch or to sight? A comparison of demographic parameter estimates obtained from mark-recapture and mark-resight models. *Biodiversity and Conservation.*23: 2781-2800.\nSoldatini, C., Y. V. Albores-Barajas, B. Massa and O. Gimenez (2014). Climate Driven Life Histories: The Case of the Mediterranean Storm Petrel. Plos One 9: e94526.\nSantin-Janin, H., B. Hugueny, P. Aubry, D. Fouchet, O. Gimenez and D. Pontier (2014). Accounting for Sampling Error When Inferring Population Synchrony from Time-Series Data: A Bayesian State-Space Modelling Approach with Applications. Plos One 9: e87084.\nHardouin, L. A., A. Robert, M. Nevoux, O. Gimenez, F. Lacroix and Y. Hingrat (2014). Meteorological conditions influence short-term survival and dispersal in a reinforced bird population. Journal of Applied Ecology. 51: 1494–1503.\nDesprez M., Harcourt R., Hindell M.A., Cubaynes S., Gimenez O., McMahon C.R. (2014). Age-specific cost of first reproduction in female southern elephant seals. Biology Letters 10: 20140264.\nCubaynes, S. C. Lavergne, O. Gimenez (2014). Fitting animal survival models with temporal random effects. Environmental and Ecological Statistics* *21: 599–610. Codes.\nLagrange, P., R. Pradel, M. Bélisle, O. Gimenez (2014). Estimating dispersal among numerous sites using capture-recapture data. Ecology. 95: 2316-2323.\nSouchay, G., O. Gimenez, G. Gauthier, R. Pradel (2014). Variations in Band Reporting Rate and Implications for Kill Rate in Greater Snow Geese. Avian Conservation and Ecology. 9: 1.\nLe Saout, S., S. Chollet, S. Chamaille-Jammes, L. Blanc, S. Padié, T. Verchère, J.A. Gaston, M. Gillingham, O. Gimenez, K.L. Parker, D. Picot, H. Verheyden, J.-L. Martin (2014). Understanding the Paradox of Deer Persisting at High Abundance in Heavily Browsed Habitats. Wildlife Biology. 20: 122-135.\nMillon, A., S.J. Petty, B. Little, O. Gimenez, T. Cornulier, X. Lambin (2014). Deleterious effect of dampening prey cycle overrides the impact of climate change on predator dynamics: a long term demographic study on tawny owls. Global Change Biology. 20: 1770–1781.\nCole, D., B.J.T. Morgan, R.S. McCrea, R. Pradel, O. Gimenez, R. Choquet (2014). Does Your Species Have Memory? Analysing Capture-Recapture Data with Memory Models. Ecology and Evolution. 4: 2124–2133.\nAbadi, F., C. Barbraud, D. Besson, J. Bried, P.-A. Crochet, K. Delord, J. Forcada, V. Grosbois, R. A. Phillipse, P. Sagar, P. Thompson, S. Waugh, H. Weimerskirch, A.G. Wood, O. Gimenez (2014). Importance of accounting for phylogenetic dependence in multi-species mark–recapture studies. Ecological Modelling. 273: 236-241.\nFrederiksen, M., Lebreton, J.-D., Pradel, R., Choquet, R. and O. Gimenez (2014). Identifying links between vital rates and environment: a toolbox for the applied ecologist. Journal of Applied Ecology. 51: 71-81.\nMarescot, L., Chapron, G., Chadès, I., Fackler, P., Duchamp, C., Marboutin, E. and O. Gimenez (2013). Complex decisions made simple: A primer on stochastic dynamic programming. Methods in Ecology and Evolution 4: 872-884. Codes.\nGamelon, M., M. Douhard, E. Baubet, O. Gimenez, S Brandt and J.-M. Gaillard (2013). Fluctuating food resources influence developmental plasticity in wild boar. Biology Letters 9: 20130419.\nDesprez, M., McMahon, C., Hindell, M., Harcourt, R. and O. Gimenez (2013). Known unknowns in an imperfect world: incorporating uncertainty in recruitment estimates using multi-event capture-recapture models. Ecology and Evolution 3: 4658-4668.\nDesprez, M., Crivelli, A.J., Lebel I., Massez G. and O. Gimenez (2013). Demographic assessment of a stocking experiment in European Eels. Ecology of Freshwater Fish 22: 412-420.\nLamy, T., O. Gimenez, J.-P. Pointier, P. Jarne and P. David (2013). Metapopulation dynamics of species with cryptic life stages. The American Naturalist 181: 479-491. Codes.\nMadon, B., C. Garrigue, R. Pradel, O. Gimenez (2013). Transience in the humpback whale population of New Caledonia and implications for abundance estimation. Marine Mammal Science 29: 669-678. Codes.\nGamelon, M., J.-M. Gaillard, E. Baubet, S. Devillard, L. Say, S. Brandt \u0026amp; O. Gimenez (2013). The relationship between phenotypic variation among offspring and mother body mass in wild boar: Evidence of coin-flipping? Journal of Animal Ecology 82: 937-945.\nChoquet, R., A. Sanz-Aguilar, B. Doligez, E. Nogué, R. Pradel, L. Gustafsson and O. Gimenez (2013). Estimating demographic parameters from capture-recapture data with dependence among individuals within clusters. Methods in Ecology and Evolution 4: 474-482.\nPéron, G., Y. Ferrand, G. Leray and O. Gimenez (2013). Waterbird demography as indicator of wetland health: The French-wintering common snipe population. Biological Conservation. 164: 123-128\nBolker, B., Gardner, B., Maunder, M., Berg, C., Brooks, M., Comita, L., Crone, E., Cubaynes, S., Davies, T., de Valpine, P., Ford, J., Gimenez, O., Kéry, M., Kim, E., Lennert-Cody, C., Magnusson, A., Martell, S., Nash, J., Nielsen, A., Regetz, J., Skaug, H., Zipkin, E. (2013). Strategies for fitting nonlinear ecological models in R, AD Model Builder, and BUGS. Methods in Ecology and Evolution 4: 501-512. Website.\nPraud, A., O. Gimenez, G. Zanella, N. Pozzi, V. Antras, L. Meyer, B. Garin-Bastuji (2013). Evaluation of five serological tests for the diagnosis of porcine brucellosis in French Polynesia. Trop Anim Health Prod 45: 933-936\nBellan, S., O. Gimenez, R. Choquet and W. Getz (2012). A Hierarchical Distance Sampling Approach to Estimating Mortality Rates from Opportunistic Carcass Surveillance Data. Methods in Ecology and Evolution 4: 361-369\nde Villemereuil, P., Gimenez, O., Doligez, B. (2012). Comparing parent-offspring regression with Frequentist and Bayesian animal models to estimate heritability in wild populations: a simulation study for Gaussian and binary traits. Methods in Ecology and Evolution 4: 260-275. Codes.\nCam, E., O. Gimenez, R. Alpizar-Jara, L. M. Aubry, M. Authier, E. G. Cooch, D. N. Koons, W. A. Link, J.-Y. Monnat, J. D. Nichols, J. J. Rotella, J. A. Royle, R. Pradel (2013). Looking for a needle in a haystack: inference about individual fitness components in a heterogeneous population. Oikos 122: 739-753.\nLebreton, J.-D. and O. Gimenez (2013). Detecting and estimating density-dependence in wildlife populations. Journal of Widlife Management 77: 12-23.\nGimenez, O., Abadi, F., Barnagaud, J.-Y., Blanc, L., Buoro, M., Cubaynes, S., Desprez, M., Gamelon, M., Guilhaumon, F., Lagrange, P., Madon, B., Marescot, L., Papadatou, E., Papaïx, J., Péron, G., Servanty, S. (2013). How can quantitative ecology be attractive to young scientists? Balancing computer/desk work with field work. Animal Conservation 16: 134-136.\nBergeron, P., Montiglio, P.-O., Reale, D., Humphries, M., Gimenez, O. and D. Garant (2013). Disruptive viability selection on adult exploratory behavior in eastern chipmunks. Journal of Evolutionary Biology 26: 766-774.\nBlanc, L., E. Marboutin, S. Gatti, O. Gimenez (2013). Abundance of rare and elusive species: empirical investigation of closed vs. spatially explicit capture-recapture models with Lynx as a case study. Journal of Wildlife Management 77: 372-378.\nPettex, E., S.-H. Lorentsen, D. Grémillet, O. Gimenez, R.T. Barrett, J.-B. Pons, C. Le Bohec and F. Bonadonna (2012). Multi-scale foraging variability in northern gannet (Morus bassanus) fuels potential foraging plasticity. Marine Biology 159: 2743-2756\nBauduin, S., J. Martin, H.H. Edwards, O. Gimenez, S. Koslovsky and D. Fagan (2012). An index of risk of co-occurrence between marine mammals and watercrafts: example of the Florida manatee. Biological Conservation 159: 127-136\nCharpentier, M.J.E., Huchard, E., Widdig, A., Gimenez, O., Sallé, B., Kappelee, P. and J.P. Renoult (2012). Distribution of affiliative behavior accross kin classes and their fitness consequences in mandrills. Ethology 118: 1198-1207\nGimenez, O., Lebreton, J.-D., Gaillard, J.-M., Choquet, R. and R. Pradel (2012). Estimating demographic parameters using hidden process dynamic models. Theoretical Population Biology 82: 307-316. Codes.\nBuoro, M., E. Prévost, O. Gimenez (2012). Digging through model complexity: using hierarchical models to uncover evolutionary processes in the wild. Journal of Evolutionary Biology 25: 2077-2090\nHardouin, L. A., M. Nevoux, A. Robert, O. Gimenez and Y. Hingrat (2012). Determinisms and costs of natal dispersal in a lekking species. Oïkos 121: 804-812\nGamelon, M., Gaillard, J.-M., S. Servanty, O. Gimenez, C. Toïgo, E. Baubet, F. Klein, J.-D. Lebreton (2012). Making use of harvest information to examine alternative management scenarios: a body weight-structured model for wild boar. Journal of Applied Ecology 49: 833-841.\nLibois, E., O. Gimenez, D. Oro, E. Mínguez, R. Pradel, A. Sanz-Aguilar (2012). Nest boxes: A successful management tool for the conservation of an endangered seabird. Biological Conservation 155: 39-43.\nAbadi, F., O. Gimenez, H. Jakobere, W. Stauberf, R. Arlettaz and M. Schaub (2012). Estimating the strength of density dependence in the presence of observation errors using integrated population models. Ecological Modelling 242: 1-9.\nLebreton, J.-D., R. Choquet and O. Gimenez (2012). Simple estimation and test procedures in capture-mark-recapture mixed models. Biometrics 68: 494-503.\nCubaynes, S., C. Lavergne, E. Marboutin, and O. Gimenez (2012). Assessing individual heterogeneity using model selection criteria: How many mixture components in capture-recapture models? Methods in Ecology and Evolution 3: 564-573. Codes.\nFrank, B., O. Gimenez and P. Baret (2012). Assessing brown trout spawning movements with multistate capture-recapture models: a case study in a fully-controlled Belgian brook. Canadian Journal of Fisheries and Aquatic Sciences 69: 1091-1104\nPéron, G., Y. Ferrand, F. Gossmann, C. Bastat, M. Guenezan and O. Gimenez (2012). Spatial heterogeneity in mortality and its impact on the population dynamics of Eurasian woodcocks. Population Ecology 54: 305-312\nCubaynes, S, C. Doutrelant, A. Gregoire, P. Perret, B. Faivre, and O. Gimenez (2012). Testing hypotheses in evolutionary ecology with imperfect detection: Structural equation modeling of mark-recapture data. Ecology 93: 248-255. Codes.\nGimenez, O., T. Anker-Nilssen and V. Grosbois (2012). Exploring causal pathways in demographic parameter variation: Path analysis of mark-recapture data. Methods in Ecology and Evolution 3: 427-432. Codes.\nChoquet, R. and O. Gimenez (2012). Towards built-in capture-recapture mixed models in program E-SURGE. Journal of Ornithology 152: 625-639.\nMarescot, L., O. Gimenez, C. Duchamp, E. Marboutin and G. Chapron (2012). Reducing matrix population models with application to social animal species. Ecological Modelling 232: 91-96\nPapadatou, E., R. Pradel, M. Schaub, D. Dolch, H. Geiger, C. Ibañez, G. Kerth, A. Popa-Lisseanu, W. Schorcht, J. Teubner and O. Gimenez (2012). Comparing survival among species with imperfect detection using multilevel analysis of mark-recapture data: a case study on bats. Ecography 35: 153-161. Codes.\nPraud, A., O. Gimenez, G. Zanella, B. Dufour, N. Pozzi, V. Antras, L. Meyer, B. Garin-Bastuji (2012). Estimation of sensitivity and specificity of five serological tests for the diagnosis of porcine brucellosis. Preventive Veterinary Medicine 104: 94-100\nArchaux, F., P.-Y. Henry and O. Gimenez (2012). When can we ignore the problem of imperfect detection in comparative studies? Methods in Ecology and Evolution 3: 188-194. Codes.\nCaniglia, R., E. Fabbri, S. Cubaynes, O. Gimenez, J.-D. Lebreton and E. Randi (2012). An improved procedure to estimate wolf abundance using non-invasive genetic sampling and capture-recapture mixture models. Conservation Genetics 13: 53-64\nFletcher, D., Lebreton, J.-D., Marescot, L., Schaub, M., Gimenez, O., Slooten, E. and S. Dawson (2012). Bias in estimation of adult survival and asymptotic population growth rate caused by undetected capture heterogeneity. Methods in Ecology and Evolution 3: 206-216\nBuoro, M., O. Gimenez and E. Prévost (2011). Assessing adaptive phenotypic plasticity by means of conditional strategies from empirical data: The latent environmental threshold model. Evolution 66: 996-1009. Codes.\nGamelon, M., A. Besnard, J.-M. Gaillard, S. Servanty, E. Baubet, S. Brandt and O. Gimenez (2011). High hunting pressure selects for earlier birth data: Wild boar as a case study. Evolution 65: 3100-3112\nMarescot, L., Pradel, R., Duchamp, C. Cubaynes, S., Marboutin, E., Choquet, R. Miquel, C. and O. Gimenez (2011). Capture-recapture population growth rate as a robust tool against detection heterogeneity for population management. Ecological Applications 21: 2898-2907\nPéron, G., Y. Ferrand, F. Gossmann, C. Bastat, M. Guenezan and O. Gimenez (2011). Escape migration decisions in Eurasian Woodcocks: insights from survival analyses using large-scale recovery data. Behavioral Ecology and Sociobiology 65: 1949-1955\nCharmantier, A., M. Buoro, O. Gimenez and H. Weimerskirch (2011). Heritability of short-scale natal dispersal in a large-scale foraging bird, the wandering albatross. Journal of Evolutionary Biology 24: 1487-1496. Codes.\nDesprez, M. R. Pradel, E. Cam, J.-Y. Monnat and O. Gimenez (2011). Now you see him, now you don\u0026rsquo;t: Experience, not age, is related to reproduction in Kittiwakes. Proceedings of the Royal Society B: Biological Sciences 278: 3060-3066.\nMadon, B., O. Gimenez, B. McArdle, C. Baker and C. Garrigue (2011). A New Method For Estimating Animal Abundance with Two Sources of Data in Capture-Recapture Studies. Methods in Ecology and Evolution 2: 390-400\nNevoux, M., O. Gimenez, M. Nicoll, D. Arlt, C. Jones and K. Norris (2011). Population regulation of territorial species: both site-dependence and interference mechanisms matter. Proceedings of the Royal Society B: Biological Sciences 278: 2173-2181\nPéron, G., Y. Ferrand, F. Gossmann, C. Bastat, M. Guenezan and O. Gimenez (2011). Nonparametric spatial regression of survival probability: visualization of population sinks in Eurasian Woodcock. Ecology 92: 1672-1679\nServanty, S., J.-M. Gaillard, Ronchi, F., Focardi, S., Baubet E. \u0026amp; O. Gimenez (2011). Influence of harvesting pressure on demographic tactics: implications for wildlife management. Journal of Applied Ecology 48: 835–843\nFontaine, R., O. Gimenez and J. Bried (2011). The impact of introduced predators, light-induced mortality of fledglings and poaching on the dynamics of the Cory’s shearwater (Calonectris diomedea) population from the Azores, northeastern subtropical Atlantic. Biological Conservation 144: 1998–2011\nMarzolin, G., A. Charmantier and O. Gimenez (2011). Frailty in state-space models: application to actuarial senescence in the dipper. Ecology 92: 562-567. Codes.\nCubaynes, S., P. F. Doherty Jr, E. A. Schreiber, R. W. Schreiber and O. Gimenez (2011). To breed or not to breed: seabirds response to extreme climatic events. Biology Letters 7: 303-306.\nPapadatou, E. Ibáñez, C., Pradel, R. Juste, J. and O. Gimenez (2011). Assessing survival in a multi-population system: a case study on bat populations. Oecologia 165: 925-933. Codes.\nBarnagaud, J.-Y., Crochet, P.-A., Magnani, Y., Bernard-Laurent, A., Menoni, E., Novoa, C. and O. Gimenez (2011). Short-term response to the North Atlantic Oscillation but no long-term effects of climate change on the reproductive success of an alpine bird. Journal of Ornithology 152: 631-641.\nMcCrea, R. S., B.J.T. Morgan, O. Gimenez, P. Besbeas, J.-D. Lebreton and T. Bregnballe (2010). Multi-site Integrated Population Modelling. Journal of Agricultural, Biological, and Environmental Statistics 15: 539-561.\nPapaïx, J., S. Cubaynes, M. Buoro, A. Charmantier, P. Perret and O. Gimenez (2010). Combining capture–recapture data and pedigree information to assess heritability of demographic parameters in the wild. Journal of Evolutionary Biology 23: 2176-2184. Codes.\nServanty, S., R. Choquet, E. Baubet, S. Brandt, J.-M. Gaillard, M. Schaub, C. Toïgo, J.-D. Lebreton, M. Buoro and O. Gimenez (2010). Assessing whether mortality is additive using marked animals: a Bayesian state-space modeling approach. Ecology 91: 1916–1923. Codes.\nBesnard, A., Novoa, C. and O. Gimenez (2010). Hunting impact on Pyrenean gray partridge population dynamics. Wildlife Biology 16: 135-143.\nBuoro, M., Prévost, E. and O. Gimenez (2010). Investigating evolutionary trade-offs in wild populations of Atlantic salmon (Salmo salar): incorporating detection probabilities and individual heterogeneity. Evolution 64: 2629-2642.\nSchaub, M. Aebischer, A. Gimenez, O., Berger, S. and R. Arlettaz (2010). Massive immigration balances high anthropogenic induced mortality in a stable eagle owl population: lessons for conservation. Biological Conservation 143: 1911–1918.\nGuilhaumon, F., Mouillot, D. and O. Gimenez (2010). mmSAR: an R-package for multimodel species-area relationship inference. Ecography 33: 420-424.\nPéron, G. Gimenez, O., Charmantier, A., Gaillard, J.-M. and P.-A. Crochet (2010). Age at the onset of senescence in birds and mammals is predicted by early-life performance. Proceedings of the Royal Society B: Biological Sciences 277: 2849–2856.\nGimenez, O. and R. Choquet (2010). Incorporating individual heterogeneity in studies on marked animals using numerical integration: capture-recapture mixed models. Ecology 91: 951-957.\nPéron, G., Crochet, P.A.C., Choquet, R., Pradel, R., Lebreton, J.-D. and O. Gimenez. (2010). Capture-recapture models with heterogeneity to study survival senescence in the wild. Oïkos 119: 524-532. Appendix.\nCubaynes, S. Pradel, R. Choquet, R. Duchamp, C. Gaillard, J.-M., Lebreton, J.-D., Marboutin, E., Miquel, C., Reboulet, A.-M., Poillot, C., Taberlet, P. and O. Gimenez. (2010). Importance of accounting for detection heterogeneity when estimating abundance: the case of French wolves. Conservation Biology 24: 621-626.\nAbadi, F., Gimenez, O., Arlettaz, R. and M. Schaub (2010). An assessment of integrated population models: bias, accuracy, and violation of the assumption of independence. Ecology 91: 7-14.\nAbadi, F., Gimenez, O., Ullrich, B., Arlettaz, R. and M. Schaub (2010). Estimation of immigration rate using integrated population models. Journal of Applied Ecology 47: 393-400.\nMihoub, J.-B., Gimenez, O., Pilard, P. and F. Sarrazin (2010). Challenging conservation of migratory species: Sahelian rainfalls drive first-year survival of the vulnerable Lesser Kestrel Falco naumanni. Biological Conservation 143: 839-847.\nGimenez, O., Grégoire, A. and T. Lenormand (2009). Estimating and visualizing fitness surfaces using mark-recapture data. Evolution. 63: 3097-3105. Codes.\nChoquet, R., Lebreton, J.-D., Gimenez, O., Reboulet, A.-M., and R. Pradel. (2009). U-CARE: Utilities for performing goodness of fit tests and manipulating CApture-REcapture data. Ecography. 32: 1071-1074.\nKazakou, E., Violle, C, Roumet, C., Pintor, C., Gimenez, O. and E. Garnier. (2009). Litter quality and decomposability of species from a Mediterranean succession depend on leaf traits but not on nitrogen supply. Annals of Botany 104: 1151-1161\nGrosbois, V., Harris, M. P., Anker-Nilssen, T., McCleery, R. H., Shaw, D. N., Morgan, B.J.T. and O. Gimenez (2009). Modeling survival at multi-population scales using mark-recapture data. Ecology 90: 2922-2932. Codes.\nGuilhaumon, F., O. Gimenez, K. J. Gaston, and D. Mouillot (2008). Reply to Bode and Murdoch: A proper integration of species-area relationship uncertainties into return on investment analyses is needed. Proceedings of the National Academy of Sciences. 106: E13\nBischof, R., J. E. Swenson, N. G. Yoccoz, A. Mysterud, and O. Gimenez (2009). The magnitude and selectivity of natural and multiple anthopogenic mortality causes in hunted brown bears. Journal of Animal Ecology 78: 656-665.\nMennerat, A., P. Perret, P. Bourgault, J. Blondel, O. Gimenez, D. W. Thomas, P. Heeb, M. M. Lambrechts (2009). Aromatic plants in nests of blue tits: positive effects on nestlings. Animal Behaviour 77: 569–574.\nGimenez, O., S. Bonner, R. King, R. A. Parker, S.P. Brooks, L. E. Jamieson, V. Grosbois, B. J. T. Morgan, and L. Thomas (2009). WinBUGS for Population Ecologists: Bayesian Modeling Using Markov Chain Monte Carlo Methods. In Modeling Demographic Processes in Marked Populations. Thomson, D. L.; Cooch, E. G.; Conroy, M. J. (Eds.). Springer Series: Environmental and Ecological Statistics, Vol. 3, pages 883-915. Codes.\nGimenez, O., Morgan, B.J.T. and S.P. Brooks (2009). Weak Identifiability in Models for Mark-Recapture-Recovery Data. In Modeling Demographic Processes in Marked Populations. Thomson, D. L.; Cooch, E. G.; Conroy, M. J. (Eds.). Springer Series: Environmental and Ecological Statistics, Vol. 3, pages 1055-1067. Codes.\nGimenez, O. and C. Barbraud (2009). The Efficient Semiparametric Regression Modeling of Capture-Recapture Data: Assessing the Impact of Climate on Survival of Two Antarctic Seabird Species. In Modeling Demographic Processes in Marked Populations. Thomson, D. L.; Cooch, E. G.; Conroy, M. J. (Eds.). Springer Series: Environmental and Ecological Statistics, Vol. 3, pages 43-58.\nBorysiewicz, R.S., B.J.T. Morgan, V. Hénaux, T. Bregnballe, J-D. Lebreton, and O. Gimenez (2009). An Integrated Analysis of Multisite Recruitment, Mark-Recapture- Recovery and Multisite Census Data. In Modeling Demographic Processes in Marked Populations. Thomson, D. L.; Cooch, E. G.; Conroy, M. J. (Eds.). Springer Series: Environmental and Ecological Statistics, Vol. 3, pages 579-591.\nGuilhaumon, F. O. Gimenez, K. J. Gaston, and D. Mouillot (2008). Taxonomic \u0026amp; regional uncertainty in species-area relationships and the identification of richness hotspots. Proceedings of the National Academy of Sciences. 105: 15458–15463.\nGimenez, O., A. Viallefont, A. Charmantier, R. Pradel, E. Cam, C. R. Brown, M. D. Anderson, M. Bomberger Brown, R. Covas, J.-M. Gaillard (2008). The risk of flawed inference in evolutionary studies when detectability is less than one. The American Naturalist. 172: 441–448.\nGimenez,O. (2008). Discussion: Towards a Bayesian analysis template? The Canadian Journal of Statistics. 36: 21-24.\nGrosbois V., Gimenez O., Gaillard, J.-M., Pradel, R., Barbraud, C., Clobert, J., Møller, A. P. and H. Weimerskirch (2008). Assessing the impact of climate variation on survival in vertebrate populations. Biological Reviews 83: 357-399.\nPradel, R., L. Maurin-Bernier, O. Gimenez, M. Genovart, R. Choquet, and D. Oro (2008). Estimation of sex-specific survival with uncertainty in sex assessment. The Canadian Journal of Statistics 36: 29-42.\nCharpentier, M.J.E., F. Prugnolle, O. Gimenez and A. Widdig (2008). Genetic heterozygosity and sociality in a primate species. Behavior Genetics 38: 151-158.\nGhimire, S., Gimenez, O., Pradel, R., McKey, D., and Y. Aumeeruddy-Thomas (2008). Demographic variation and population viability in a threatened Himalayan medicinal and aromatic herb (Nardostachys grandiflora): matrix modelling of harvesting effects in two contrasting habitats. Journal of Applied Ecology 45: 41-51.\nVéran, S., Gimenez, O., Flint, E., Kendall, B., Doherty, P., and J.-D. Lebreton (2007). Quantifying the impact of longline fisheries on adult survival in the Black-footed Albatross. Journal of Applied Ecology 44: 942–952.\nGimenez, O., V. Rossi, R. Choquet, C. Dehais, B. Doris, H. Varella, J.-P. Vila and R. Pradel (2007). State-space modelling of data on marked individuals. Ecological Modelling 206: 431-438. Codes.\nSchaub, M., Gimenez, O., Sierro, A. and R. Arlettaz (2007). Use of Integrated Modeling to Enhance Estimates of Population Dynamics Obtained from Limited Data. Conservation Biology 21: 945–955. Codes.\nKazakou, E., E. Garnier and Gimenez, O. (2007). Contribution of leaf lifespan and nutrient resorption to mean residence time: an elasticity analysis. Ecology 88: 1857-1863.\nPinay, G., B. Gumiero, E. Tabacchi, O. Gimenez, A.M. Tabacchi-Planty, M.M. Hefting, T.P. Burt, V.A. Black, C. Nilsson, V. Iordache, F. Bureau, L. Vought, G.E. Petts and H. Décamps. (2007). Patterns of denitrification rates in European alluvial soils under various hydrological regimes. Freshwater Biology 52: 252–266.\nGimenez O., C. Crainiceanu, C. Barbraud, S. Jenouvrier and B.J.T. Morgan. (2006). Semiparametric regression in capture-recapture modelling. Biometrics 62: 691-698. Codes.\nGimenez, O., R. Covas, C. R. Brown, M. D. Anderson, M. Bomberger Brown and T. Lenormand. (2006). Nonparametric estimation of natural selection on a quantitative trait using mark-recapture data. Evolution 60: 460-466. Codes.\nCharpentier M., Peignot P., Hossaert-McKey M., Gimenez O., Setchell J. and E.J. Wickings (2005). Constraints on control: factors influencing reproductive success in male mandrils (Mandrillus sphinx). Behavioral Ecology 16: 614-623.\nDo F.C., Goudiaby V.A., Gimenez O., Diagne A.L., Diouf M., Rocheteau A., Akpo L.E. (2005) Environmental influence on canopy phenology in the dry tropics Forest Ecology and Management. 215: 319-328.\nGimenez O., Choquet R., Lamor L., Scofield P., Fletcher D., J.-D. Lebreton and R. Pradel (2005). Efficient Profile-Likelihood Confidence Intervals for Capture-Recapture Models. Journal of Agricultural, Biological, and Environmental Statistics 10: 184-196.\nPradel, R., Gimenez O. and J.-D. Lebreton (2005). Principles and interest of GOF tests for multistate capture-recapture models. Animal Biodiversity and Conservation 28: 189–204.\nAncrenaz M., Gimenez O., Ambu L., Ancrenaz K., Andau P., Goossens B., Payne J., Tuuga A. and I. Lackman- Ancrenaz (2005). Aerial Survey Gives New Estimates for Orang-utans in Sabah, Malaysia. Plos Biology 3: 30-37.\nGimenez O., Viallefont A., Choquet R., Catchpole E.A. and B.J.T. Morgan (2004). Methods for investigating parameter redundancy. Animal Biodiversity and Conservation 27: 561-572.\nChoquet R., Reboulet A.-M., Pradel R., Gimenez O. and J.-D. Lebreton (2004). M-SURGE: new software specically designed for multistate capture-recapture models. Animal Biodiversity and Conservation 27: 207-215.\nSchaub M., Gimenez O., Schmidt B.R. and R. Pradel (2004). Estimating survival and temporary emigration in the multistate capture-recapture framework. Ecology 85: 2107-2113. Codes.\nAncrenaz M., Goossens B., Gimenez O., Sawang A. and I. Lackman-Ancrenaz (2004). Determination of ape distribution and population size using ground and aerial surveys: a case study with orang-utans in lower Kinabatangan, Sabah, Malaysia. Animal Conservation 7: 375-385.\nCharpentier M., Hossaert-McKey M., Peignot P., Gimenez O., Setchell J.M. and E.J. Wickings (2004). Social factors and male reproductive success in a highly sexually dimorphic primate: Paternity analysis over a 20 year period in a group of mandrills (Mandrillus sphinx). Folia Primatologica 75: 244-245.\nGimenez O., Choquet R. and J.-D. Lebreton (2003). Parameter redundancy in multistate capture-recapture models. Biometrical Journal 45: 704-722. Codes.\nPradel R., Wintrebert C.M.A. and Gimenez O. (2003). A proposal for a goodness-of-fit test to the Arnason-Schwarz multisite capture-recapture model. Biometrics 59: 43-53.\nBesnard A., Gimenez O. and J.-D. Lebreton (2002). A model for the evolution of creching behaviour in gulls. Evolutionary Ecology 16: 489-503.\n","date":1697846400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1697846400,"objectID":"79a6ff5fd591560441afe05143df5626","permalink":"https://oliviergimenez.github.io/publication/papers/","publishdate":"2023-10-21T00:00:00Z","relpermalink":"/publication/papers/","section":"publication","summary":"Preprints\nMoulherat S., L. Pautrel, G. Debat, M.-P. Etienne, L. Gendron, N. Hautière, J.-P. Tarel, G. Testud, O. Gimenez (2023). Biodiversity Monitoring with Intelligent Sensors: An Integrated Pipeline for Mitigating Animal-Vehicle Collisions.","tags":null,"title":"Papers","type":"publication"},{"authors":null,"categories":null,"content":"July 12-13, 2023. Workshop on Bayesian capture-recapture inference with hidden Markov models in R and Nimble for the València International Bayesian Analysis Summer School. Slides, worksheets and video recordings freely available from dedicated website.\nJune 1-2, 2023. Quantitative methods for population dynamics: A two-day workshop with R. All material available from dedicated website.\nMay 16, 2023. Workshop on estimating (wolf) occupancy with R for the Wolves Across Borders conference. All material available from dedicated website.\nJune 26, 2022. Workshop on Bayesian capture-recapture inference with hidden Markov models, R and Nimble for the ISEC2022 conference. All material available from dedicated website.\nMarch 21-22, 2022. Quantitative methods for population dynamics: A two-day workshop with R. All material available from dedicated website.\nMarch 19, 2022. Workshop on NIMBLE. All material available from dedicated website.\nJune 21-22, 2021. Workshop on individual-based models with R. All material available from dedicated website.\nMay 17-18, 2021. Workshop on Bayesian capture-recapture inference with hidden Markov models in R and Nimble. Slides, worksheets and video recordings freely available from dedicated website.\nMarch 22-26, 2021. Bayesian statistics with R for non-stats PhD students. Slides and videos here.\nMarch 18, 2021. Workshop on reproducible science in R and RStudio. Remotely via Zoom. Slides and videos here.\nNovember 23-27, 2020. Integrated population models. Remotely via Zoom. Slides and videos on capture-recapture models here.\nNovember 2020. Bonnes pratiques pour une recherche reproductible en écologie numérique. Montpellier, France.\nOctober 2020. Statistics for ecologists Master course.\nJuly 2020. Bayesian statistics with R and Jags. Remotely via Zoom.\nDécembre 2019. Bonnes pratiques pour une recherche reproductible en écologie numérique. Montpellier, France.\n2010-2018. Multievent capture-recapture models. Montpellier, France.\n","date":1689984000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1689984000,"objectID":"4139306fa16fab1c8cb885cd90f0622d","permalink":"https://oliviergimenez.github.io/talks/workshop/","publishdate":"2023-07-22T00:00:00Z","relpermalink":"/talks/workshop/","section":"talks","summary":"July 12-13, 2023. Workshop on Bayesian capture-recapture inference with hidden Markov models in R and Nimble for the València International Bayesian Analysis Summer School. Slides, worksheets and video recordings freely available from dedicated website.","tags":null,"title":"Workshops","type":"talks"},{"authors":null,"categories":null,"content":" Peyrard N. and Gimenez O. (2022). Statistical Approaches for Hidden Variables in Ecology. La version française est ici. Newman, K.B., S.T. Buckland, B.J.T. Morgan, R. King, D.L. Borchers, D.J. Cole, P. Besbeas, O. Gimenez and L. Thomas (2014). Modelling Population Dynamics - Model Formulation, Fitting and Assessment using State-Space Methods. King, R., B.J.T. Morgan, O. Gimenez, S.P. Brooks (2009). Bayesian Analysis of Population Ecology. Morrissey, M.B., P. de Villemereuil, B. Doligez, O. Gimenez (2014). Bayesian approaches to the quantitative genetic analysis of natural populations. In Quantitative Genetics in the Wild, Editors: A. Charmantier, D. Garant, and L.E.B. Kruuk. Oxford University Press, pages 228-253. Grosbois V. and O. Gimenez (2010). Capture-mark-recapture models. In Effects of Climate Change on Birds, Editors: A.P. Moller, W. Fiedler and P. Berthold, Oxford University Press, pages 39-46. Lebreton, J.-D., E. Klein, O. Gimenez and F. Rousset (2010). Statistiques pour la biologie évolutive. Dans Biologie Evolutive, Editeurs: Thomas F, Lefèvre T \u0026amp; Raymond M. De Boeck Editions, pages 33-48. Rouan L., S. Cubaynes, C. Duchamp, C. Miquel, A.-.M. Reboulet, O. Gimenez, J.-D. Lebreton, R. Choquet and R. Pradel (2010). Modèles de mélange en capture-recapture. Dans Modèles à variables latentes et modèles de mélange, Editeurs: G. Saporta, J.J. Droesbeke and C. Thomas-Agnan, Editions Technip, pages 265-272. ","date":1666569600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1666569600,"objectID":"e6b9c30b973c4d01c41005b0bd7d4026","permalink":"https://oliviergimenez.github.io/publication/books/","publishdate":"2022-10-24T00:00:00Z","relpermalink":"/publication/books/","section":"publication","summary":"Peyrard N. and Gimenez O. (2022). Statistical Approaches for Hidden Variables in Ecology. La version française est ici. Newman, K.B., S.T. Buckland, B.J.T. Morgan, R. King, D.L. Borchers, D.J. Cole, P.","tags":null,"title":"Books and book chapters","type":"publication"},{"authors":null,"categories":null,"content":"Un point d\u0026#39;étape sur nos projets sur la loutre d\u0026#39;Europe. J\u0026#39;y parle de connectivité, trame turquoise, pièges photos, ADN env, cacas, représentations des usagers et médiation scientifique. Et de caca, j\u0026#39;ai déjà dit caca? 🦦💩🧬📸\nPhotos : Yann Raulethttps://t.co/5CakMhkggi pic.twitter.com/pkhaVQwPOV\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) December 9, 2023 ","date":1702080000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1702080000,"objectID":"0c5ca14563a824b30703979501f3515f","permalink":"https://oliviergimenez.github.io/blog/updateotterconnect/","publishdate":"2023-12-09T00:00:00Z","relpermalink":"/blog/updateotterconnect/","section":"blog","summary":"Un point d\u0026#39;étape sur nos projets sur la loutre d\u0026#39;Europe. J\u0026#39;y parle de connectivité, trame turquoise, pièges photos, ADN env, cacas, représentations des usagers et médiation scientifique. Et de caca, j\u0026#39;ai déjà dit caca?","tags":["statistical ecology","Otter","scientific outreach"],"title":"Update on our Otter projects","type":"blog"},{"authors":null,"categories":null,"content":"Petit exposé sur les modèles et leurs applications en écologie à l\u0026#39;occasion d\u0026#39;un atelier d\u0026#39;échanges Modélisations « Biodiversité et Infrastructures Linéaires de Transport » organisé par @CeremaCom @OFBiodiversite @CentreTVB @Ecologie_Gouv #ITTECOP https://t.co/tLOWezfc69 pic.twitter.com/QjlOelYSMU\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) November 28, 2023 ","date":1701129600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1701129600,"objectID":"19fc9e1393fe13c1f29bff2316b519c7","permalink":"https://oliviergimenez.github.io/blog/modelscerema/","publishdate":"2023-11-28T00:00:00Z","relpermalink":"/blog/modelscerema/","section":"blog","summary":"Petit exposé sur les modèles et leurs applications en écologie à l\u0026#39;occasion d\u0026#39;un atelier d\u0026#39;échanges Modélisations « Biodiversité et Infrastructures Linéaires de Transport » organisé par @CeremaCom @OFBiodiversite @CentreTVB @Ecologie_Gouv #ITTECOP https://t.","tags":["statistical ecology","models","wolf"],"title":"Models and applications in ecology","type":"blog"},{"authors":null,"categories":null,"content":"La team loutre est sur place ! Venez vous balader au festival des sciences participatives et citoyennes https://t.co/N6I94VU2iJ#VaSavoir @instantscience @cefemontpellier @CNRSecologie @CNRS_OccitaniE @montpellier_ 🦦https://t.co/CHEJnrW9KE pic.twitter.com/tSe2uSnHsG\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) October 28, 2023 ","date":1698451200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1698451200,"objectID":"2bc6b33a1b6a004c08ed3260b5ace586","permalink":"https://oliviergimenez.github.io/blog/instantscience/","publishdate":"2023-10-28T00:00:00Z","relpermalink":"/blog/instantscience/","section":"blog","summary":"La team loutre est sur place ! Venez vous balader au festival des sciences participatives et citoyennes https://t.co/N6I94VU2iJ#VaSavoir @instantscience @cefemontpellier @CNRSecologie @CNRS_OccitaniE @montpellier_ 🦦https://t.co/CHEJnrW9KE pic.twitter.com/tSe2uSnHsG\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) October 28, 2023 ","tags":["scientific outreach","citizen science","eurasian otters"],"title":"Otters and citizen science","type":"blog"},{"authors":null,"categories":null,"content":"Excited to be running a workshop on Bayesian capture-recapture inference w hidden Markov models, #rstats \u0026amp; @R_nimble 🥳\nThanks for the invitation @vibass7 🥰\nWebsite is up and running w slides, code \u0026amp; data https://t.co/X7MYs8ec5X pic.twitter.com/UDzIVeVhuZ\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) July 10, 2023 ","date":1688947200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1688947200,"objectID":"b861076dda9838d52138ed8ae9cbda13","permalink":"https://oliviergimenez.github.io/blog/vibassworkshop/","publishdate":"2023-07-10T00:00:00Z","relpermalink":"/blog/vibassworkshop/","section":"blog","summary":"Excited to be running a workshop on Bayesian capture-recapture inference w hidden Markov models, #rstats \u0026amp; @R_nimble 🥳\nThanks for the invitation @vibass7 🥰\nWebsite is up and running w slides, code \u0026amp; data https://t.","tags":["workshop","capture-recapture","nimble","bayes","rstats"],"title":"Workshop on Bayesian analysis of capture-recapture data with hidden Markov models","type":"blog"},{"authors":null,"categories":null,"content":"Au festival #AgropolEat une sortie organisée avec mon collègue Vincent Sablain #EPTBLez pour sensibiliser à la présence de la loutre d\u0026#39;Europe sur le Lez 🦦💩📸🧬\nBonne nouvelle pour la biodiversité, de nouveaux équilibres à construire\nInfos 👉 https://t.co/CHEJnrW9KE\nVidéos ⤵️ pic.twitter.com/E1nwD2xRJM\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) July 2, 2023 ","date":1688256000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1688256000,"objectID":"80175ccb86ece5813d407b12d613f3fa","permalink":"https://oliviergimenez.github.io/blog/agropoleat/","publishdate":"2023-07-02T00:00:00Z","relpermalink":"/blog/agropoleat/","section":"blog","summary":"Au festival #AgropolEat une sortie organisée avec mon collègue Vincent Sablain #EPTBLez pour sensibiliser à la présence de la loutre d\u0026#39;Europe sur le Lez 🦦💩📸🧬\nBonne nouvelle pour la biodiversité, de nouveaux équilibres à construire","tags":["festival","OtterConnect","vulgarisation"],"title":"La loutre au festival Agropol'Eat","type":"blog"},{"authors":null,"categories":null,"content":"Notre labo @cefemontpellier a eu la visite du PDG @CNRS \u0026amp; du président @umontpellier, à qui j\u0026#39;ai présenté nos travaux sur les gds carnivores \u0026amp; l\u0026#39;écologie statistique 🐺🔍🧮@antoine_petit_ @Philippe_Auge @NavasLaure @jerome_vitre @OFBiodiversite 🧵▶️ https://t.co/IIV3aX2nKo pic.twitter.com/JR252ix8rV\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) June 20, 2023 ","date":1687219200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1687219200,"objectID":"cf450f6d85de55227be42c422d106b17","permalink":"https://oliviergimenez.github.io/blog/visitepdg/","publishdate":"2023-06-20T00:00:00Z","relpermalink":"/blog/visitepdg/","section":"blog","summary":"Notre labo @cefemontpellier a eu la visite du PDG @CNRS \u0026amp; du président @umontpellier, à qui j\u0026#39;ai présenté nos travaux sur les gds carnivores \u0026amp; l\u0026#39;écologie statistique 🐺🔍🧮@antoine_petit_ @Philippe_Auge @NavasLaure @jerome_vitre @OFBiodiversite 🧵▶️ https://t.","tags":["CNRS"],"title":"Visite du PDG du CNRS au CEFE","type":"blog"},{"authors":null,"categories":null,"content":"Really proud of this new paper in @AER_ESE_BES in which we give our two cents on how to fill the gap bw researchers \u0026amp; managers w/ our long-term collab bw our lab @cefemontpellier \u0026amp; the French Biodiversity Agency @OFBiodiversite https://t.co/JrMcpFr9yl@CNRS_OccitaniE 🧵👇 pic.twitter.com/D7RO6VZA1W\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) June 7, 2023 ","date":1686096000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1686096000,"objectID":"b000ee28e5ece73efb033542d3a05ad5","permalink":"https://oliviergimenez.github.io/blog/newpaperofb/","publishdate":"2023-06-07T00:00:00Z","relpermalink":"/blog/newpaperofb/","section":"blog","summary":"Really proud of this new paper in @AER_ESE_BES in which we give our two cents on how to fill the gap bw researchers \u0026amp; managers w/ our long-term collab bw our lab @cefemontpellier \u0026amp; the French Biodiversity Agency @OFBiodiversite https://t.","tags":["research management gap","OFB","CNRS"],"title":"New paper alert","type":"blog"},{"authors":null,"categories":null,"content":"Pour Occitanie Est, la semaine \u0026quot;écologie environnement biodiversité\u0026quot; aura lieu le 13 juin 14h30 et on y parlera des travaux des jeunes chercheurs et chercheuses, et des collaborations avec la ville/métro de Montpellier. Inscriptions jusqu\u0026#39;à vendredi ▶️ https://t.co/6Hw2BAvGAV https://t.co/mWwGICvaqZ pic.twitter.com/4ryMBcnhKL\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) June 7, 2023 ","date":1686096000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1686096000,"objectID":"c835359b48ec765b923533c1d37aaad3","permalink":"https://oliviergimenez.github.io/blog/semaineinee2023/","publishdate":"2023-06-07T00:00:00Z","relpermalink":"/blog/semaineinee2023/","section":"blog","summary":"Pour Occitanie Est, la semaine \u0026quot;écologie environnement biodiversité\u0026quot; aura lieu le 13 juin 14h30 et on y parlera des travaux des jeunes chercheurs et chercheuses, et des collaborations avec la ville/métro de Montpellier.","tags":["conférence","biodiversité","CNRS","INEE"],"title":"Semaine écologie environnement biodiversité du CNRS","type":"blog"},{"authors":null,"categories":null,"content":"🥳🤩😇 yet another great paper by @ValentinLauret from his PhD w/ Hélène Labach, Léa David and Matthieu Authier 😍\nthx @masonfidino @PaulCaplat 😉@cefemontpellier @CNRS_OccitaniE @INEE_CNRS @MIRACETI_ceta @UnivLaRochelle @umontpellier https://t.co/GT7pdMXauN\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) June 2, 2023 ","date":1685664000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1685664000,"objectID":"6cf0cbbf820d2feaf73a83080887db99","permalink":"https://oliviergimenez.github.io/blog/newpaperval/","publishdate":"2023-06-02T00:00:00Z","relpermalink":"/blog/newpaperval/","section":"blog","summary":"🥳🤩😇 yet another great paper by @ValentinLauret from his PhD w/ Hélène Labach, Léa David and Matthieu Authier 😍\nthx @masonfidino @PaulCaplat 😉@cefemontpellier @CNRS_OccitaniE @INEE_CNRS @MIRACETI_ceta @UnivLaRochelle @umontpellier https://t.co/GT7pdMXauN\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) June 2, 2023 ","tags":["multispecies occupancy","bottlenose dolphins","photo-id"],"title":"New paper alert","type":"blog"},{"authors":null,"categories":null,"content":"Aaaaaand let\u0026#39;s start this thing 😉 @abesnardEPHE @SarahCubaynes @javi_ferlop @ValentinLauret @MarwanNaciri2 @ThibautCouturie @LiseViollat @cthierrycc\n▶️ https://t.co/jlTE4inuHM@cefemontpellier @INEE_CNRS @CNRS_OccitaniE @EPHE_PSL @umontpellier https://t.co/Iehm6DdhjM\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) June 1, 2023 ","date":1685577600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1685577600,"objectID":"4befbd8d289aca8ecb979b9cd7dea38d","permalink":"https://oliviergimenez.github.io/blog/workshoppopdyn2023/","publishdate":"2023-06-01T00:00:00Z","relpermalink":"/blog/workshoppopdyn2023/","section":"blog","summary":"Aaaaaand let\u0026#39;s start this thing 😉 @abesnardEPHE @SarahCubaynes @javi_ferlop @ValentinLauret @MarwanNaciri2 @ThibautCouturie @LiseViollat @cthierrycc\n▶️ https://t.co/jlTE4inuHM@cefemontpellier @INEE_CNRS @CNRS_OccitaniE @EPHE_PSL @umontpellier https://t.co/Iehm6DdhjM\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) June 1, 2023 ","tags":["workshop","population dynamics","rstats","team work"],"title":"Workshop on population dynamics","type":"blog"},{"authors":null,"categories":null,"content":"It\u0026#39;s #WorldOtterDay and we\u0026#39;re happy to share our research on the Eurasian otter and socio-ecological connectivity using cam traps, eDNA, spraints, interviews #interdisciplinarity\nCheck out https://t.co/QwwJtY2Aeq for info @IOSF @montpellier_ @Montpellier3m @umontpellier 🎥👇 pic.twitter.com/dfphlFdwY6\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) May 31, 2023 ","date":1685491200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1685491200,"objectID":"4cde6f12900e63743f6e69e410b082dc","permalink":"https://oliviergimenez.github.io/blog/worldotterday/","publishdate":"2023-05-31T00:00:00Z","relpermalink":"/blog/worldotterday/","section":"blog","summary":"It\u0026#39;s #WorldOtterDay and we\u0026#39;re happy to share our research on the Eurasian otter and socio-ecological connectivity using cam traps, eDNA, spraints, interviews #interdisciplinarity\nCheck out https://t.co/QwwJtY2Aeq for info @IOSF @montpellier_ @Montpellier3m @umontpellier 🎥👇 pic.","tags":["eurasian otter","OtterConnect"],"title":"World Otter Day","type":"blog"},{"authors":null,"categories":null,"content":"🚨 New paper 🚨 by @PatriRodrigs testing the \u0026quot;humans as shield\u0026quot; \u0026amp; \u0026quot;humans as super-predators\u0026quot; hypotheses w/ camp trap data on mammals using multispecies occupancy models à la Rota et al. 🤩😇https://t.co/NMlm3loxG8 pic.twitter.com/P4upBAzb8g\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) May 30, 2023 ","date":1685404800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1685404800,"objectID":"688eacab394dc33981f65d3fd8531b67","permalink":"https://oliviergimenez.github.io/blog/newpaperpat/","publishdate":"2023-05-30T00:00:00Z","relpermalink":"/blog/newpaperpat/","section":"blog","summary":"🚨 New paper 🚨 by @PatriRodrigs testing the \u0026quot;humans as shield\u0026quot; \u0026amp; \u0026quot;humans as super-predators\u0026quot; hypotheses w/ camp trap data on mammals using multispecies occupancy models à la Rota et al.","tags":["multispecies occupancy","human shield hypothesis","camera traps"],"title":"New paper alert","type":"blog"},{"authors":null,"categories":null,"content":"New preprint led by Marine Desprez w/ Vincent Miele on \u0026quot;Nine tips for ecologists using machine learning\u0026quot; 🤩#MachineLearning #Ecology #RStats #preprinthttps://t.co/KvXTmhABtF@LbbeLyon @cefemontpellier @INEE_CNRS @CNRS_OccitaniE @umontpellier pic.twitter.com/CbiFOIjXhX\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) May 22, 2023 ","date":1684713600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1684713600,"objectID":"238aecd74817a3e86678b9a9ed7ba7ce","permalink":"https://oliviergimenez.github.io/blog/preprintml/","publishdate":"2023-05-22T00:00:00Z","relpermalink":"/blog/preprintml/","section":"blog","summary":"New preprint led by Marine Desprez w/ Vincent Miele on \u0026quot;Nine tips for ecologists using machine learning\u0026quot; 🤩#MachineLearning #Ecology #RStats #preprinthttps://t.co/KvXTmhABtF@LbbeLyon @cefemontpellier @INEE_CNRS @CNRS_OccitaniE @umontpellier pic.twitter.com/CbiFOIjXhX\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) May 22, 2023 ","tags":["machine learning","computer vision","tips","deep learning","artificial intelligence","rstats"],"title":"Nine tips for ecologists using machine learning","type":"blog"},{"authors":null,"categories":null,"content":"I just run a 3h workshop on estimating occupancy in R w/ wolf as a case study 🐺🗺️ 🌍 Website w/ slides, R code and data https://t.co/KOKaPRPRtv\n📽️ Video recording (in English w/ 👽 accent) https://t.co/LlmkLtrd8L#WolvesBorders @WolvesBorders #RStats 🧵👇 pic.twitter.com/yz8gb0Hc3C\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) May 16, 2023 ","date":1684195200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1684195200,"objectID":"5af3a6e4d07ac236b808779dd6226dc7","permalink":"https://oliviergimenez.github.io/blog/workshopoccupancy/","publishdate":"2023-05-16T00:00:00Z","relpermalink":"/blog/workshopoccupancy/","section":"blog","summary":"I just run a 3h workshop on estimating occupancy in R w/ wolf as a case study 🐺🗺️ 🌍 Website w/ slides, R code and data https://t.co/KOKaPRPRtv\n📽️ Video recording (in English w/ 👽 accent) https://t.","tags":["statistical ecology","occupancy","rstats","unmarked"],"title":"Workshop on estimating occupancy in R","type":"blog"},{"authors":null,"categories":null,"content":"Le Réseau Thématique en écologie statistique propose un annuaire des labos/pers en #ecostat Que vous soyez à la recherche d\u0026#39;un labo pour un stage, d\u0026#39;une pers avec une compétence, ou pour vous faire une idée des contours de notre communauté, visitez https://t.co/jaXdqPFVKe 🤩🥳 pic.twitter.com/qSrS7XHrso\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) May 5, 2023 ","date":1683244800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1683244800,"objectID":"d743300396898aa9d5e1125d253e9003","permalink":"https://oliviergimenez.github.io/blog/networkgdrecostat/","publishdate":"2023-05-05T00:00:00Z","relpermalink":"/blog/networkgdrecostat/","section":"blog","summary":"Le Réseau Thématique en écologie statistique propose un annuaire des labos/pers en #ecostat Que vous soyez à la recherche d\u0026#39;un labo pour un stage, d\u0026#39;une pers avec une compétence, ou pour vous faire une idée des contours de notre communauté, visitez https://t.","tags":["statistical ecology","network"],"title":"Annuaire du Réseau Thématique en écologie statistique","type":"blog"},{"authors":null,"categories":["project"],"content":"Context. Conservation and the sustainable use of biodiversity and ecosystem services largely depend on the successful management of ecological populations, which in turn depend on our knowledge and our ability to understand and predict how the structure and dynamics of ecological populations respond to changes in their environments.\nRationale and aim. Despite its importance, the quantification of the effects of anthropic pressures on biodiversity is difficult and rarely undertaken because of the lack of species-specific ecological data and/or human pressures and activities, and the complexity and lack of transfer of analytical methods needed to evaluate these long-term consequences. The DISCAR (DIS = disturbance, CAR = carnivores) project aims at providing an operational framework with analytical tools for assessing the impacts or population consequences of human pressures, and showcasing it with case studies in applied conservation.\nCase study. In DISCAR, we will use small carnivores in French mainland and overseas territories as a case study to assess the population consequences or impacts of human pressures on animal populations. Why small carnivores? Because these species hold important roles in ecosystems, such as influencing ecosystem structure and providing numerous ecosystem services, including pest/disease control and seed dispersal. Small carnivore species also respond to ecosystem perturbation in a fast, measurable, and interpretable way and therefore can be used as sentinels of global change.\nObjectives. DISCAR will connect the fields of bioenergetics and demographic modeling in an integrated framework to assess human pressures on biodiversity, and to characterize then predict the impacts on small carnivores. DISCAR will identify key parameters underlying coupling between pressures and impacts, and will make recommendations for stakeholders in terms of monitoring and mitigation strategies.\nWork packages. We will i) assemble a unique dataset on small carnivores including their ecology, bioenergetic requirements, demography and human pressures, ii) quantify and characterize the impacts of interacting pressures on populations, and iii) forecast the fate of populations to assess their viability under various scenarios of pressures and mitigation to help devise sound conservation strategies.\nConsortium. To achieve these objectives, we have gathered international experts in ecological modelling and small carnivores who will develop and transfer an analytical framework for quantification, characterization and prediction of the impacts of pressures on populations. These experts are:\nOlivier Gimenez – CNRS (France) Sandrine Ruette – OFB (France) Courtney Marneweck – Giraffe Conservation Foundation (South Africa) Patrick Giraudoux – Université Bourgogne Franche-Comté (France) Isabel Smallegange - Newcastle University (UK) Sébastien Devillard – Université Claude Bernard Lyon 1 (France) Cara Gallagher - University of Potsdam (Allemagne) Sarah Cubaynes – EPHE (France) John Fryxell – University of Guelph (Canada) Elizabeth McHuron – University of Washington (USA) Rob Salguero-Gomez – University of Oxford (UK) Marie-Lazarine Poulle – Université Reims Champagne Ardenne (France) Funding. This project is funded by the French Foundation for Biodiversity Research (FRB) through Cesab.\n","date":1683072000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1683072000,"objectID":"40cfee82f289cf02f742bd4383ac7165","permalink":"https://oliviergimenez.github.io/my-project/external-project/discar/","publishdate":"2023-05-03T00:00:00Z","relpermalink":"/my-project/external-project/discar/","section":"my-project","summary":"Population consequences of human DISturbance on small CARnivores.","tags":["small carnivores","monitoring","management"],"title":"DISCAR","type":"my-project"},{"authors":null,"categories":["project"],"content":"","date":1683072000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1683072000,"objectID":"1a1d94cebff1799227ca00b4ec8a8395","permalink":"https://oliviergimenez.github.io/my-project/external-project/otters/","publishdate":"2023-05-03T00:00:00Z","relpermalink":"/my-project/external-project/otters/","section":"my-project","summary":"An applied research project to better understand otter recolonization.","tags":["carnivores","monitoring","eDNA","cam-traps"],"title":"OtterConnect","type":"my-project"},{"authors":null,"categories":null,"content":"Last week we had the pleasure to host the #euring2023 conf 🤩 Awesome pple, amazing science, sun \u0026amp; good food 😇 Hope everyone enjoyed it 💜\nVideo recordings of talks \u0026amp; workshops freely available at https://t.co/aPG6PXHw6Q PDF at https://t.co/i3Ew3sbe9h\n🧵👇 w/ many 🙏 1/15\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) April 24, 2023 ","date":1682294400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1682294400,"objectID":"5ec6e01115fffa541132ce28f774c719","permalink":"https://oliviergimenez.github.io/blog/euring2023/","publishdate":"2023-04-24T00:00:00Z","relpermalink":"/blog/euring2023/","section":"blog","summary":"Last week we had the pleasure to host the #euring2023 conf 🤩 Awesome pple, amazing science, sun \u0026amp; good food 😇 Hope everyone enjoyed it 💜\nVideo recordings of talks \u0026amp; workshops freely available at https://t.","tags":["statistical ecology","euring","conference"],"title":"Euring 2023","type":"blog"},{"authors":null,"categories":null,"content":"Jetez un coup d\u0026rsquo;oeil à https://youtu.be/xbUoXrEsDV4.\n😇🥳 Super reportage @F3OccitanieLR sur notre projet loutre OtterConnect https://t.co/CHEJnrW9KE avec Vincent Sablain #EPTBLez 🦦💩📸🧬 Soutenu par @montpellier_ @Montpellier3m @umontpellier @zoobeauval @cefemontpellier ▶️ 8\u0026#39;36\u0026#39;\u0026#39; https://t.co/JFp8dNJhxc\n🙏 @SebastienBanus\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) April 24, 2023 ","date":1682294400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1682294400,"objectID":"ee4dacdb28664d2d248030001cc11e66","permalink":"https://oliviergimenez.github.io/blog/reportagefrance3/","publishdate":"2023-04-24T00:00:00Z","relpermalink":"/blog/reportagefrance3/","section":"blog","summary":"Jetez un coup d\u0026rsquo;oeil à https://youtu.be/xbUoXrEsDV4.\n😇🥳 Super reportage @F3OccitanieLR sur notre projet loutre OtterConnect https://t.co/CHEJnrW9KE avec Vincent Sablain #EPTBLez 🦦💩📸🧬 Soutenu par @montpellier_ @Montpellier3m @umontpellier @zoobeauval @cefemontpellier ▶️ 8\u0026#39;36\u0026#39;\u0026#39; https://t.","tags":["otter","media","reportage","otterconnect"],"title":"Reportage France 3 sur notre projet loutre OtterConnect","type":"blog"},{"authors":null,"categories":null,"content":"Oksana Grente @oksanagrente a fait sa thèse sur les déprédations par le loup et le contrôle létal 🐺🐑 Pour porter ses résultats à connaissance de toutes et tous, elle a écrit cet article https://t.co/UaBgeWdySu. C\u0026#39;est en français, et plus court que le manuscrit de \u0026gt; 300 pages 😉 https://t.co/HgeoLlYe3P pic.twitter.com/tNDw8pIEzm\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) April 6, 2023 ","date":1680739200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1680739200,"objectID":"d8751db68a97561b6e725d9fe0cf6962","permalink":"https://oliviergimenez.github.io/blog/naturae/","publishdate":"2023-04-06T00:00:00Z","relpermalink":"/blog/naturae/","section":"blog","summary":"Oksana Grente @oksanagrente a fait sa thèse sur les déprédations par le loup et le contrôle létal 🐺🐑 Pour porter ses résultats à connaissance de toutes et tous, elle a écrit cet article https://t.","tags":["wildlife management","wolf","paper"],"title":"déprédations par le loup et contrôle létal","type":"blog"},{"authors":null,"categories":null,"content":"I\u0026rsquo;m starting a new project on the Eurasian otter to better understand its recolonization in France.\nIn the OtterConnect project, we aim at providing an operational framework for assessing socio-ecological connectivity for small carnivores which influence the dynamics of ecosystems and provide numerous ecosystem services. We focus on the Eurasian otter (Lutra lutra) an umbrella and indicator species of blue and green infrastructures. To know more about the project and get news, please visit our website https://otterconnect.netlify.app/en/.\n","date":1680480000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1680480000,"objectID":"1aa8b6063edf649f82c39891754ce034","permalink":"https://oliviergimenez.github.io/blog/otterconnect/","publishdate":"2023-04-03T00:00:00Z","relpermalink":"/blog/otterconnect/","section":"blog","summary":"I\u0026rsquo;m starting a new project on the Eurasian otter to better understand its recolonization in France.\n","tags":["small carnivores","eDNA","cam-traps","monitoring"],"title":"A new project on the Eurasian otter and connectivity","type":"blog"},{"authors":null,"categories":null,"content":"How to assess landscape connectivity with spatial capture-recapture models? @oSCR_package to the rescue ! Some code to reproduce results from a great paper by Dana Morin @Fuller_Lab @andyroyle_pwrc \u0026amp; @chrissuthy Post: https://t.co/FVwBzMnqA9\nCode: https://t.co/FMMyF0s22x pic.twitter.com/QUlOjlGnPf\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) March 27, 2023 ","date":1679875200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1679875200,"objectID":"19aa9591e6698352b661513d1a604187","permalink":"https://oliviergimenez.github.io/blog/connectoscr/","publishdate":"2023-03-27T00:00:00Z","relpermalink":"/blog/connectoscr/","section":"blog","summary":"How to assess landscape connectivity with spatial capture-recapture models? @oSCR_package to the rescue ! Some code to reproduce results from a great paper by Dana Morin @Fuller_Lab @andyroyle_pwrc \u0026amp; @chrissuthy Post: https://t.","tags":["connectivity","R","rstats"],"title":"How to assess landscape connectivity with spatial capture-recapture models?","type":"blog"},{"authors":null,"categories":null,"content":"📢👋 Together with @SarahCubaynes \u0026amp; @abesnardEPHE we will give a 2-day introductory workshop June 1-2 on quantitative methods for population dynamics in R #rstats Join us 🥳 It\u0026#39;s online and free of charge 😉 You just need to register via https://t.co/otguiSMTY0\nPlease RT 😇 pic.twitter.com/6ghDHE92SJ\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) February 8, 2023 ","date":1675814400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1675814400,"objectID":"8eefd2814c017810c2eb7201267c404a","permalink":"https://oliviergimenez.github.io/blog/annonceworkshoppopdyn/","publishdate":"2023-02-08T00:00:00Z","relpermalink":"/blog/annonceworkshoppopdyn/","section":"blog","summary":"📢👋 Together with @SarahCubaynes \u0026amp; @abesnardEPHE we will give a 2-day introductory workshop June 1-2 on quantitative methods for population dynamics in R #rstats Join us 🥳 It\u0026#39;s online and free of charge 😉 You just need to register via https://t.","tags":["population dynamics","workshop","R","rstats"],"title":"Workshop on quantitative methods for population dynamics in R","type":"blog"},{"authors":null,"categories":null,"content":"Ey! VIBASS 6 is announced online now https://t.co/VS8cLR8ocK\n🗓️ July 10-14 2023\n🚩Faculty of Mathematics @FMatemaUV_EG, Valencia, Spain Be aware that the registration would start really really soon and we just have few spots. Don\u0026#39;t miss it. pic.twitter.com/5M61dbuxuF\n\u0026mdash; VIBASS 6 (@vibass7) January 13, 2023 ","date":1673827200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1673827200,"objectID":"6618d54253a8831cb977ce9bb200d9db","permalink":"https://oliviergimenez.github.io/blog/annoncevibass/","publishdate":"2023-01-16T00:00:00Z","relpermalink":"/blog/annoncevibass/","section":"blog","summary":"Ey! VIBASS 6 is announced online now https://t.co/VS8cLR8ocK\n🗓️ July 10-14 2023\n🚩Faculty of Mathematics @FMatemaUV_EG, Valencia, Spain Be aware that the registration would start really really soon and we just have few spots.","tags":["capture-recapture","hmm","workshop","bayesian statistics","R","rstats"],"title":"VIBASS and Bayesian capture-recapture workshop","type":"blog"},{"authors":null,"categories":null,"content":"Check it out, this is a cool ++++ paper by @MaelisKervellec 🤩🤩 So glad for her that she got her first PhD paper published #OpenAccess @OFBiodiversite @CNRS_OccitaniE @INEE_CNRS @umontpellier @twitthair1 @cefemontpellier https://t.co/sVdXl40DyU\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) January 11, 2023 ","date":1673395200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1673395200,"objectID":"36368e03349387c2981d323096bc747b","permalink":"https://oliviergimenez.github.io/blog/papiermaelis/","publishdate":"2023-01-11T00:00:00Z","relpermalink":"/blog/papiermaelis/","section":"blog","summary":"Check it out, this is a cool ++++ paper by @MaelisKervellec 🤩🤩 So glad for her that she got her first PhD paper published #OpenAccess @OFBiodiversite @CNRS_OccitaniE @INEE_CNRS @umontpellier @twitthair1 @cefemontpellier https://t.","tags":["spatial capture-recapture","connectivity","brown bear"],"title":"Maëlis Kervellec's first PhD paper!","type":"blog"},{"authors":null,"categories":null,"content":"This morning we\u0026#39;ll have a panorama of the math/stat/info models we use/develop at @cefemontpellier to address questions in ecology \u0026amp; evolution 🤩😇 Program below. This is an internal seminar to get to know each other better, but @Master_RAINET students will be live tweeting 😍🙏 https://t.co/qp6qYMGlOh pic.twitter.com/xrSXIIBZhm\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) December 6, 2022 ","date":1670284800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1670284800,"objectID":"ea89b7fba946012f6c8d9a2b647bd4bc","permalink":"https://oliviergimenez.github.io/blog/modelocefe/","publishdate":"2022-12-06T00:00:00Z","relpermalink":"/blog/modelocefe/","section":"blog","summary":"This morning we\u0026#39;ll have a panorama of the math/stat/info models we use/develop at @cefemontpellier to address questions in ecology \u0026amp; evolution 🤩😇 Program below. This is an internal seminar to get to know each other better, but @Master_RAINET students will be live tweeting 😍🙏 https://t.","tags":["modelling"],"title":"Modelling in ecology and evolution at CEFE","type":"blog"},{"authors":null,"categories":null,"content":"So proud of our latest paper on estimating abundance of Brown bear in the Pyrenees 🐻📈 in @PeerComJournal led by C Vanpé 😇https://t.co/aBZtWmO1UW\nYet another example of our collaboration w/ @OFBiodiversite 🤩\nData \u0026amp; code https://t.co/uZGQ3Fy2Iq#DiamondOpenAccess pic.twitter.com/H0PfGWFFxD\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) November 21, 2022 ","date":1668988800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1668988800,"objectID":"6fab3b283002d7f043648e5164f1dbd4","permalink":"https://oliviergimenez.github.io/blog/bearsize/","publishdate":"2022-11-21T00:00:00Z","relpermalink":"/blog/bearsize/","section":"blog","summary":"So proud of our latest paper on estimating abundance of Brown bear in the Pyrenees 🐻📈 in @PeerComJournal led by C Vanpé 😇https://t.co/aBZtWmO1UW\nYet another example of our collaboration w/ @OFBiodiversite 🤩","tags":["brown bear","capture-recapture"],"title":"Brown bear abundance in the Pyrenees mountains","type":"blog"},{"authors":null,"categories":null,"content":"Conditions de travail et santé mentale des jeunes chercheur.e.s : constats et solutions. La vidéo de la conf d\u0026#39;Adèle Combes @Vies_de_these et de la table ronde qui a suivi est dispo sur https://t.co/Sv4nT7zMmE. https://t.co/A9UH0aX5Un\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) November 8, 2022 ","date":1667865600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1667865600,"objectID":"45b60aded2cbe4c118a142681c6479f9","permalink":"https://oliviergimenez.github.io/blog/adele/","publishdate":"2022-11-08T00:00:00Z","relpermalink":"/blog/adele/","section":"blog","summary":"Conditions de travail et santé mentale des jeunes chercheur.e.s : constats et solutions. La vidéo de la conf d\u0026#39;Adèle Combes @Vies_de_these et de la table ronde qui a suivi est dispo sur https://t.","tags":["mental health","academic life"],"title":"Conference on mental health and working conditions in academia","type":"blog"},{"authors":null,"categories":null,"content":".@LouvrierJulie reçoit le prix jeune chercheuse 2020 de la @sfecologie au @Salon_AdNatura #AdNatura 🥳 pour son travail de thèse sur le loup 🐺🗺 @cefemontpellier @CNRS_OccitaniE @INEE_CNRS @umontpellier @OFBiodiversite pic.twitter.com/T2dNxeuaA9\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) October 29, 2022 ","date":1667001600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1667001600,"objectID":"8e69bae4ada6ec03b0a809f906b7091c","permalink":"https://oliviergimenez.github.io/blog/juliesfe/","publishdate":"2022-10-29T00:00:00Z","relpermalink":"/blog/juliesfe/","section":"blog","summary":".@LouvrierJulie reçoit le prix jeune chercheuse 2020 de la @sfecologie au @Salon_AdNatura #AdNatura 🥳 pour son travail de thèse sur le loup 🐺🗺 @cefemontpellier @CNRS_OccitaniE @INEE_CNRS @umontpellier @OFBiodiversite pic.twitter.com/T2dNxeuaA9\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) October 29, 2022 ","tags":["PhD","French society for ecology and evolution"],"title":"Julie Louvrier reçoit le prix jeune chercheuse 2020 de la SFEE","type":"blog"},{"authors":null,"categories":null,"content":"💜💜💜 https://t.co/YWC2qSepIr\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) October 25, 2022 ","date":1666656000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1666656000,"objectID":"8e6c85c37d6704ca14c56553d0cdd2cd","permalink":"https://oliviergimenez.github.io/blog/teamlunch/","publishdate":"2022-10-25T00:00:00Z","relpermalink":"/blog/teamlunch/","section":"blog","summary":"💜💜💜 https://t.co/YWC2qSepIr\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) October 25, 2022 ","tags":["Hair team","team building"],"title":"Team lunch!","type":"blog"},{"authors":null,"categories":null,"content":"Check it out 😉🥳 https://t.co/mqyuMSm8QF\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) September 22, 2022 ","date":1663804800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1663804800,"objectID":"d3ef175cddc6396f245d4ef533ced606","permalink":"https://oliviergimenez.github.io/blog/valpaper/","publishdate":"2022-09-22T00:00:00Z","relpermalink":"/blog/valpaper/","section":"blog","summary":"Check it out 😉🥳 https://t.co/mqyuMSm8QF\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) September 22, 2022 ","tags":["integrated modelling","nimble"],"title":"Integrating distance sampling and spatial capture-recapture to assess density of bottlenose dolphins","type":"blog"},{"authors":null,"categories":null,"content":"Looking for free images or icons for your talk, website, video, etc. A few pointers that I find useful 😉https://t.co/oTioaam8Y1https://t.co/pueZRTCONFhttps://t.co/nhx3wlPHhvhttps://t.co/6XT1z2jgg5https://t.co/2KCJU3xCAPhttps://t.co/QGnYaLsSjmhttps://t.co/TVlAtmlVDz pic.twitter.com/SoUshEWAgM\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) August 18, 2022 ","date":1660780800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1660780800,"objectID":"209998c5b52eb947c81f2edf9673b5cd","permalink":"https://oliviergimenez.github.io/blog/icons/","publishdate":"2022-08-18T00:00:00Z","relpermalink":"/blog/icons/","section":"blog","summary":"Looking for free images or icons for your talk, website, video, etc. A few pointers that I find useful 😉https://t.co/oTioaam8Y1https://t.co/pueZRTCONFhttps://t.co/nhx3wlPHhvhttps://t.co/6XT1z2jgg5https://t.co/2KCJU3xCAPhttps://t.co/QGnYaLsSjmhttps://t.co/TVlAtmlVDz pic.twitter.com/SoUshEWAgM\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) August 18, 2022 ","tags":["resources","icons"],"title":"Resources on finding images and icons","type":"blog"},{"authors":null,"categories":null,"content":"How do neural networks work 🤔 What\u0026#39;s under the hood 🤯 A few pointers in the thread below with a R/statistian bias 😇 #MachineLearning #DeepLearning #NeuralNetworks #CNN #RStats 🧵⬇️ pic.twitter.com/4QF1vKoFc8\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) August 8, 2022 ","date":1659916800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1659916800,"objectID":"c3cd4ff80fdecb3816212ca467bd2b00","permalink":"https://oliviergimenez.github.io/blog/neuralnetworks/","publishdate":"2022-08-08T00:00:00Z","relpermalink":"/blog/neuralnetworks/","section":"blog","summary":"How do neural networks work 🤔 What\u0026#39;s under the hood 🤯 A few pointers in the thread below with a R/statistian bias 😇 #MachineLearning #DeepLearning #NeuralNetworks #CNN #RStats 🧵⬇️ pic.twitter.com/4QF1vKoFc8","tags":["neural networks","statistical ecology"],"title":"Resources on neural networks for ecology","type":"blog"},{"authors":null,"categories":null,"content":"I’m writing a lot over summer. Writing does not come easily to me. Experience helps of course, but I find it useful to follow some rules, like how to write an intro. Fortunately, the best writing tips are out there 😉\nMy fav 🧵👇@OpenAcademics#AcademicTwitter#AcademicChatter pic.twitter.com/v2CeZKF6jx\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) July 27, 2022 ","date":1658880000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1658880000,"objectID":"ea1fdb47e27f65f85bdfc26de7cdf4b1","permalink":"https://oliviergimenez.github.io/blog/write/","publishdate":"2022-07-27T00:00:00Z","relpermalink":"/blog/write/","section":"blog","summary":"I’m writing a lot over summer. Writing does not come easily to me. Experience helps of course, but I find it useful to follow some rules, like how to write an intro.","tags":["writing","academic life"],"title":"Writing tips","type":"blog"},{"authors":null,"categories":null,"content":"C\u0026#39;est la période de rédaction de thèse pour bcp de jeunes chercheuses/chercheurs. Une étape pas facile à passer. Un fil sur comment on s\u0026#39;y prend dans l\u0026#39;équipe. Ça vaut ce que ça vaut, c\u0026#39;est pas original, mais si ça peut aider, tant mieux.\nEt vous, quels sont vos conseils?\n🧵⬇️ pic.twitter.com/p8qw4ZoDuQ\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) July 17, 2022 ","date":1658016000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1658016000,"objectID":"067b7c73faf0b58923c5fa0caffe97f3","permalink":"https://oliviergimenez.github.io/blog/advicephdwriting/","publishdate":"2022-07-17T00:00:00Z","relpermalink":"/blog/advicephdwriting/","section":"blog","summary":"C\u0026#39;est la période de rédaction de thèse pour bcp de jeunes chercheuses/chercheurs. Une étape pas facile à passer. Un fil sur comment on s\u0026#39;y prend dans l\u0026#39;équipe. Ça vaut ce que ça vaut, c\u0026#39;est pas original, mais si ça peut aider, tant mieux.","tags":["PhD","academic life"],"title":"Conseils pour écrire sa thèse","type":"blog"},{"authors":[],"categories":[],"content":"If you need to test the goodness-of-fit of your open capture-recapture models, good news, our #RStats R2ucare is back on CRAN thanks to the amazing work of Murray Efford and @bc_stevenson 🙏🤩\nwebsite: https://t.co/SbGrQm9XS2 paper: https://t.co/W38nqsEHPa pic.twitter.com/KelIzsB8Xf\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 12, 2022 ","date":1657670400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1657670400,"objectID":"dd07f490d66f126a7fe5df3639cd794e","permalink":"https://oliviergimenez.github.io/authors/2022-07-13-nos-travaux-sur-le-loup/","publishdate":"2022-07-13T00:00:00Z","relpermalink":"/authors/2022-07-13-nos-travaux-sur-le-loup/","section":"authors","summary":"If you need to test the goodness-of-fit of your open capture-recapture models, good news, our #RStats R2ucare is back on CRAN thanks to the amazing work of Murray Efford and @bc_stevenson 🙏🤩","tags":["loup"],"title":"Nos travaux sur le loup","type":"authors"},{"authors":[],"categories":[],"content":"Notre équipe @twitthair1 travaille bcp sur les questions en lien avec la gestion du loup en France 🐺🇫🇷 Voici qqs pointeurs sur nos travaux, en privilégiant les productions en français. Plusieurs de ces résultats ont été obtenus avec les collègues @OFBiodiversite 🧵⬇️\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 13, 2022 ","date":1657670400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1657670400,"objectID":"f7c7277e1cdf2a9d65a5558ad08a927e","permalink":"https://oliviergimenez.github.io/blog/2022-07-13-nos-travaux-sur-le-loup/","publishdate":"2022-07-13T00:00:00Z","relpermalink":"/blog/2022-07-13-nos-travaux-sur-le-loup/","section":"blog","summary":"Notre équipe @twitthair1 travaille bcp sur les questions en lien avec la gestion du loup en France 🐺🇫🇷 Voici qqs pointeurs sur nos travaux, en privilégiant les productions en français. Plusieurs de ces résultats ont été obtenus avec les collègues @OFBiodiversite 🧵⬇️","tags":["loup"],"title":"Nos travaux sur le loup","type":"blog"},{"authors":[],"categories":[],"content":"If you need to test the goodness-of-fit of your open capture-recapture models, good news, our #RStats R2ucare is back on CRAN thanks to the amazing work of Murray Efford and @bc_stevenson 🙏🤩\nwebsite: https://t.co/SbGrQm9XS2 paper: https://t.co/W38nqsEHPa pic.twitter.com/KelIzsB8Xf\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 12, 2022 ","date":1657584000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1657584000,"objectID":"294bec081990d9e0c8c03a35f410baf2","permalink":"https://oliviergimenez.github.io/blog/2022-07-12-r2ucare-back-on-cran/","publishdate":"2022-07-12T00:00:00Z","relpermalink":"/blog/2022-07-12-r2ucare-back-on-cran/","section":"blog","summary":"If you need to test the goodness-of-fit of your open capture-recapture models, good news, our #RStats R2ucare is back on CRAN thanks to the amazing work of Murray Efford and @bc_stevenson 🙏🤩","tags":["R2ucare","package","ANR"],"title":"R2ucare back on CRAN","type":"blog"},{"authors":[],"categories":[],"content":"🥳 Save the date 🥳\nThe 2023 Euring Analytical Conference will be held in Montpellier (France), on April 17-21, 2023.\nCheck out https://t.co/t04XJGbW8k for info. 🤗🤩We look forward to welcoming you all 🤗🤩@_EURING @abesnardEPHE pic.twitter.com/KAe46wOaze\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 8, 2022 ","date":1657238400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1657238400,"objectID":"b086efcbffbddc67888bc35a939c83da","permalink":"https://oliviergimenez.github.io/blog/2022-07-08-save-the-data-euring2023-in-montpellier-april-17-21/","publishdate":"2022-07-08T00:00:00Z","relpermalink":"/blog/2022-07-08-save-the-data-euring2023-in-montpellier-april-17-21/","section":"blog","summary":"🥳 Save the date 🥳\nThe 2023 Euring Analytical Conference will be held in Montpellier (France), on April 17-21, 2023.\nCheck out https://t.co/t04XJGbW8k for info. 🤗🤩We look forward to welcoming you all 🤗🤩@_EURING @abesnardEPHE pic.","tags":["euring","capture-recapture","conference"],"title":"Save the data: EURING2023 in Montpellier April 17-21","type":"blog"},{"authors":[],"categories":[],"content":"Quick \u0026amp; dirty text mining analysis of #ISEC2022 abstracts using topic modelling. Nothing out of the ordinary. Would be curious to assess time trends in topics w/ longitudinal analysis @chrissuthy 😇😉\nData and code: https://t.co/E2x7avw0E1\nPaper: https://t.co/l3ChAIa6Vi pic.twitter.com/NUF1FGm3Lv\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 4, 2022 ","date":1656892800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1656892800,"objectID":"13387cf60284fc9cdcff4ea39409f6cd","permalink":"https://oliviergimenez.github.io/blog/2022-07-04-quick-and-dirty-text-mining-analysis-of-isec2022-abstracts/","publishdate":"2022-07-04T00:00:00Z","relpermalink":"/blog/2022-07-04-quick-and-dirty-text-mining-analysis-of-isec2022-abstracts/","section":"blog","summary":"Quick \u0026amp; dirty text mining analysis of #ISEC2022 abstracts using topic modelling. Nothing out of the ordinary. Would be curious to assess time trends in topics w/ longitudinal analysis @chrissuthy 😇😉","tags":[],"title":"Quick and dirty text mining analysis of ISEC2022 abstracts","type":"blog"},{"authors":[],"categories":[],"content":"Would you like to get the most out of your capture-recapture data?\nWe run a workshop on multisite/state/event HMMs models in #Rstats Nimble w/ Daniel Turek @R_nimble at #ISEC2022 @ISEC_stats_ecol\nWebsite w/ slides \u0026amp; codes: https://t.co/gI1ccZO3O9\nVideo: https://t.co/zA6nh6Ttih pic.twitter.com/ejyE2tCw5o\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 1, 2022 ","date":1656633600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1656633600,"objectID":"0d30c499b67b1892f4dc7dcbd9a91251","permalink":"https://oliviergimenez.github.io/blog/2022-07-01-workshop-at-isec2022-on-capture-recapture-hmms-and-nimble/","publishdate":"2022-07-01T00:00:00Z","relpermalink":"/blog/2022-07-01-workshop-at-isec2022-on-capture-recapture-hmms-and-nimble/","section":"blog","summary":"Would you like to get the most out of your capture-recapture data?\nWe run a workshop on multisite/state/event HMMs models in #Rstats Nimble w/ Daniel Turek @R_nimble at #ISEC2022 @ISEC_stats_ecol","tags":["nimble","capture-recapture","hideen-markov-models","hmm"],"title":"Workshop at ISEC2022 on capture-recapture HMMs and Nimble","type":"blog"},{"authors":[],"categories":[],"content":"Thrilled to see our work on Mediterranean monk seal out in @AnimalConserv 🦭\nWe show a significant range recovery of the species in Greece 🇬🇷 using #CitizenScience \u0026amp; occupancy modelshttps://t.co/ILYdzArZq7\nCheck out NGO MOm amazing work https://t.co/hM8BhfRQ7w\nDM ➡️ PDF 😉 pic.twitter.com/ILqnFvMypL\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 30, 2022 ","date":1656547200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1656547200,"objectID":"74c4442d5b4312cf358991d71cb78136","permalink":"https://oliviergimenez.github.io/blog/2022-06-30-new-paper-on-monk-seal-recovery-in-greece/","publishdate":"2022-06-30T00:00:00Z","relpermalink":"/blog/2022-06-30-new-paper-on-monk-seal-recovery-in-greece/","section":"blog","summary":"Thrilled to see our work on Mediterranean monk seal out in @AnimalConserv 🦭\nWe show a significant range recovery of the species in Greece 🇬🇷 using #CitizenScience \u0026amp; occupancy modelshttps://t.co/ILYdzArZq7","tags":["dynamic-occupancy","monk-seals","citizen-science"],"title":"New paper on monk seal recovery in Greece","type":"blog"},{"authors":[],"categories":[],"content":"I\u0026#39;m always reluctant to give career advice, cause what worked for me won\u0026#39;t work for others, and there is obviously a survivorship bias. For what it\u0026#39;s worth, my 10 pieces of advice to early career researchers in statistical ecology #ISEC2022 🙏 @BayesRuth for the Q 😉\n🧵👇 https://t.co/JJEHFFOAc9\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 28, 2022 ","date":1656460800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1656460800,"objectID":"610f48360d290731338b120bc6ff1da0","permalink":"https://oliviergimenez.github.io/blog/2022-06-29-career-advice-to-early-career-researchers/","publishdate":"2022-06-29T00:00:00Z","relpermalink":"/blog/2022-06-29-career-advice-to-early-career-researchers/","section":"blog","summary":"I\u0026#39;m always reluctant to give career advice, cause what worked for me won\u0026#39;t work for others, and there is obviously a survivorship bias. For what it\u0026#39;s worth, my 10 pieces of advice to early career researchers in statistical ecology #ISEC2022 🙏 @BayesRuth for the Q 😉","tags":["ECR","academic-life"],"title":"Career advice to Early Career Researchers","type":"blog"},{"authors":[],"categories":[],"content":"Statistical ecology 🌍📊, hidden Markov models 🛰️ and large carnivores 🐺😺🐻 Soooo happy to share our research in this plenary talk at #ISEC2022 @ISEC_stats_ecol 🤩🍾\nslides and text: https://t.co/BOtZHWU56L\nvideo: https://t.co/6Qn7i05uae pic.twitter.com/vrSYeo9HYF\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 28, 2022 ","date":1656374400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1656374400,"objectID":"3d001a111f047e215e4d4f48e36c6d77","permalink":"https://oliviergimenez.github.io/blog/2022-06-28-my-plenary-talk-at-isec2022-on-statistical-ecology-hidden-markov-models-and-large-carnivores/","publishdate":"2022-06-28T00:00:00Z","relpermalink":"/blog/2022-06-28-my-plenary-talk-at-isec2022-on-statistical-ecology-hidden-markov-models-and-large-carnivores/","section":"blog","summary":"Statistical ecology 🌍📊, hidden Markov models 🛰️ and large carnivores 🐺😺🐻 Soooo happy to share our research in this plenary talk at #ISEC2022 @ISEC_stats_ecol 🤩🍾\nslides and text: https://t.co/BOtZHWU56L\nvideo: https://t.","tags":["hidden-markov-models","hmm","dynamic-occupancy","capture-recapture","conference"],"title":"My plenary talk at ISEC2022 on Statistical ecology, hidden Markov models and large carnivores.","type":"blog"},{"authors":[],"categories":[],"content":"Yesterday I created my 200th repo on @github 😇 It\u0026#39;s been quite an adventure to twist my old habits \u0026amp; switch to #ReproducibleResearch but it\u0026#39;s definitely worth it 🤯 Plenty of help out there like @living_norway 🤩 If useful, I also have a 1-day workshop https://t.co/7I6rU61nEg pic.twitter.com/cGDjaA6iTv\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 9, 2022 ","date":1654732800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1654732800,"objectID":"c52fe494c94078f74b0928f0afb551ed","permalink":"https://oliviergimenez.github.io/blog/2022-06-09-reproducible-research-and-github/","publishdate":"2022-06-09T00:00:00Z","relpermalink":"/blog/2022-06-09-reproducible-research-and-github/","section":"blog","summary":"Yesterday I created my 200th repo on @github 😇 It\u0026#39;s been quite an adventure to twist my old habits \u0026amp; switch to #ReproducibleResearch but it\u0026#39;s definitely worth it 🤯 Plenty of help out there like @living_norway 🤩 If useful, I also have a 1-day workshop https://t.","tags":["github","git","reproducible-research"],"title":"Reproducible research and GitHub","type":"blog"},{"authors":[],"categories":[],"content":"Check it out, this is huge 🤯 Awesome initiative led by @FredBarraquand Let\u0026#39;s get ourselves a 💎 #OpenAccess journal in #StatisticalEcology 🤩 Free for readers and for authors https://t.co/HmDhrCLsaN https://t.co/AI7Jjunoek pic.twitter.com/Aywk2ucusQ\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 8, 2022 ","date":1654646400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1654646400,"objectID":"f06e40d805610dbcbe32dcc7a0258141","permalink":"https://oliviergimenez.github.io/blog/2022-06-08-a-proposal-for-a-diamond-open-access-journal-in-statistical-ecology/","publishdate":"2022-06-08T00:00:00Z","relpermalink":"/blog/2022-06-08-a-proposal-for-a-diamond-open-access-journal-in-statistical-ecology/","section":"blog","summary":"Check it out, this is huge 🤯 Awesome initiative led by @FredBarraquand Let\u0026#39;s get ourselves a 💎 #OpenAccess journal in #StatisticalEcology 🤩 Free for readers and for authors https://t.co/HmDhrCLsaN https://t.co/AI7Jjunoek pic.","tags":["open-access","open-science","statistical-ecology"],"title":"A proposal for a diamond Open Access journal in Statistical Ecology","type":"blog"},{"authors":[],"categories":[],"content":"Because of my old age I guess, I\u0026#39;m often asked how I switched from base R #RStats to #tidyverse. I learnt by watching @drob screencasts 🎥 in which he analyses #TidyTuesday data he has never seen before (!). Here is a list of my tips https://t.co/VFAaDz8onE in case it\u0026#39;s useful. pic.twitter.com/4Emq2J2qQL\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 1, 2022 ","date":1654041600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1654041600,"objectID":"357e908f6c1ad3f64bd6631590cfc614","permalink":"https://oliviergimenez.github.io/blog/2022-06-01-tips-and-tricks-in-tidyverse/","publishdate":"2022-06-01T00:00:00Z","relpermalink":"/blog/2022-06-01-tips-and-tricks-in-tidyverse/","section":"blog","summary":"Because of my old age I guess, I\u0026#39;m often asked how I switched from base R #RStats to #tidyverse. I learnt by watching @drob screencasts 🎥 in which he analyses #TidyTuesday data he has never seen before (!","tags":["tidyverse","dplyr","ggplot"],"title":"Tips and tricks in Tidyverse","type":"blog"},{"authors":[],"categories":[],"content":"A Montpellier, @oaggimenez présente le fonctionnement du DIPEE Occitanie-Est qui rassemble plusieurs laboratoires @UmrAmap @cefemontpellier @criobe_pf @isemevol @umrMARBEC @MIVEGEC @ihpe_lab @EcotronCNRS Pour en savoir plus sur ce DIPEE 👇https://t.co/QvGniMkjg2 pic.twitter.com/7ekUx3vwu5\n\u0026mdash; L\u0026#39;écologie au CNRS (@INEE_CNRS) May 31, 2022 ","date":1653955200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1653955200,"objectID":"7b7075e6832a56c2276e329554b853fc","permalink":"https://oliviergimenez.github.io/blog/2022-05-31-semaine-ecologie-environnement-biodiversite-du-cnrs-a-montpellier/","publishdate":"2022-05-31T00:00:00Z","relpermalink":"/blog/2022-05-31-semaine-ecologie-environnement-biodiversite-du-cnrs-a-montpellier/","section":"blog","summary":"A Montpellier, @oaggimenez présente le fonctionnement du DIPEE Occitanie-Est qui rassemble plusieurs laboratoires @UmrAmap @cefemontpellier @criobe_pf @isemevol @umrMARBEC @MIVEGEC @ihpe_lab @EcotronCNRS Pour en savoir plus sur ce DIPEE 👇https://t.co/QvGniMkjg2 pic.twitter.com/7ekUx3vwu5","tags":["cnrs","conference"],"title":"Semaine écologie environnement biodiversité du CNRS à Montpellier","type":"blog"},{"authors":[],"categories":[],"content":"It\u0026#39;s real 😇🤩🥳 https://t.co/HdWol2H6PJ pic.twitter.com/NUSXtS1fQy\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) May 24, 2022 ","date":1653350400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1653350400,"objectID":"8a6d3af2659a8384cefa8de9b4a567ef","permalink":"https://oliviergimenez.github.io/blog/2022-05-24-our-new-book-on-statistical-approaches-for-hiddens-variables-in-ecology/","publishdate":"2022-05-24T00:00:00Z","relpermalink":"/blog/2022-05-24-our-new-book-on-statistical-approaches-for-hiddens-variables-in-ecology/","section":"blog","summary":"It\u0026#39;s real 😇🤩🥳 https://t.co/HdWol2H6PJ pic.twitter.com/NUSXtS1fQy\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) May 24, 2022 ","tags":["hmm","hidden-markov-models","latent-variables","state-space-models","statistical-ecology"],"title":"Our new book on statistical approaches for hiddens variables in ecology","type":"blog"},{"authors":[],"categories":[],"content":"🔧🚧 I have updated my quick introduction to data visualisation w/ R 📦 ggplot2 🥳\nIn case it\u0026#39;s useful 😉▶️ https://t.co/IWNesVjwqq ◀️\nw/ practicals https://t.co/YYkykGeWf1 \u0026amp; solutions https://t.co/NCQYsw8QCg Cartoons @allison_horst 💜#ggplot #tidyverse #dataviz #RStats pic.twitter.com/RlXFT9F0Tw\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) May 23, 2022 ","date":1653264000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1653264000,"objectID":"e1e6487655975f75170253a1eb37d2ea","permalink":"https://oliviergimenez.github.io/blog/2022-05-23-my-quick-introduction-to-data-visualisation-with-r/","publishdate":"2022-05-23T00:00:00Z","relpermalink":"/blog/2022-05-23-my-quick-introduction-to-data-visualisation-with-r/","section":"blog","summary":"🔧🚧 I have updated my quick introduction to data visualisation w/ R 📦 ggplot2 🥳\nIn case it\u0026#39;s useful 😉▶️ https://t.co/IWNesVjwqq ◀️\nw/ practicals https://t.co/YYkykGeWf1 \u0026amp; solutions https://t.co/NCQYsw8QCg Cartoons @allison_horst 💜#ggplot #tidyverse #dataviz #RStats pic.","tags":["ggplot","dataviz","data-visualisation"],"title":"My quick introduction to data visualisation with R","type":"blog"},{"authors":[],"categories":[],"content":"🔧🚧 Intro on data wrangling w/ R 📦 dplyr - import data, pipe, pivot, join, filter, count, work w/ columns, dates, char, factors, etc #tidyverse\n▶️ https://t.co/Vs2TbYhbBj ◀️\nw/ practicals https://t.co/9ktKPp9wfW \u0026amp; solutions https://t.co/YPwibL2w31 Cartoons @allison_horst 💜 pic.twitter.com/ZMSfxYFKWq\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) May 17, 2022 ","date":1652745600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1652745600,"objectID":"904a99402fa0b6816cd97671aee28761","permalink":"https://oliviergimenez.github.io/blog/2022-05-17-introduction-on-data-wrangling-with-r/","publishdate":"2022-05-17T00:00:00Z","relpermalink":"/blog/2022-05-17-introduction-on-data-wrangling-with-r/","section":"blog","summary":"🔧🚧 Intro on data wrangling w/ R 📦 dplyr - import data, pipe, pivot, join, filter, count, work w/ columns, dates, char, factors, etc #tidyverse\n▶️ https://t.co/Vs2TbYhbBj ◀️\nw/ practicals https://t.","tags":["dplyr","R","data-wrangling"],"title":"Introduction on data wrangling with R","type":"blog"},{"authors":[],"categories":[],"content":"😇 I updated the slides of my introduction to #GIS and #mapping in #rstats using the #sf package and 🐻 🗺️ in the #pyrenees 🏔️ 🧑‍🏫 Slides: https://t.co/mO4Dg8l1H5\n🧑‍💻 Code: https://t.co/X8fDKcrbFR#rspatial #spatial #ggplot2 #tidyverse #dataviz pic.twitter.com/mSsL6Qt7IS\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) May 12, 2022 ","date":1652313600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1652313600,"objectID":"b1711fff361df7eef2af6caa4f9a0663","permalink":"https://oliviergimenez.github.io/blog/2022-05-12-introduction-to-gis-and-mapping-in-r/","publishdate":"2022-05-12T00:00:00Z","relpermalink":"/blog/2022-05-12-introduction-to-gis-and-mapping-in-r/","section":"blog","summary":"😇 I updated the slides of my introduction to #GIS and #mapping in #rstats using the #sf package and 🐻 🗺️ in the #pyrenees 🏔️ 🧑‍🏫 Slides: https://t.co/mO4Dg8l1H5\n🧑‍💻 Code: https://t.","tags":["sf","R","gis","sig"],"title":"Introduction to GIS and mapping in R","type":"blog"},{"authors":[],"categories":[],"content":"Happy to share lecture slides \u0026amp; R codes of an earlier version of the workshop 👇 we ran back in 2014 🤯 w JD Lebreton \u0026amp; @KoonsLab ➡️ https://t.co/XdcCQ560Pz #RStats For video 📽️📺, check out our workshop on pop dynamics https://t.co/jlTE4iF5zk w @SarahCubaynes \u0026amp; @abesnardEPHE https://t.co/JFgS1GOU0V\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) May 4, 2022 ","date":1651622400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1651622400,"objectID":"47c5f57a24998da50317c378568d6dc0","permalink":"https://oliviergimenez.github.io/blog/2022-05-04-matrix-population-models-lecture-slides-and-r-codes-for-a-workshop/","publishdate":"2022-05-04T00:00:00Z","relpermalink":"/blog/2022-05-04-matrix-population-models-lecture-slides-and-r-codes-for-a-workshop/","section":"blog","summary":"Happy to share lecture slides \u0026amp; R codes of an earlier version of the workshop 👇 we ran back in 2014 🤯 w JD Lebreton \u0026amp; @KoonsLab ➡️ https://t.co/XdcCQ560Pz #RStats For video 📽️📺, check out our workshop on pop dynamics https://t.","tags":["matrix-population-models","caswell","R","open-science"],"title":"Matrix population models: Lecture slides and R codes for a workshop","type":"blog"},{"authors":[],"categories":[],"content":"Useful papers on GLMMs 🤩🥳@bolkerb https://t.co/HX41N05rwK@BobOHara https://t.co/c8fVvxG7ru@mattjsilk https://t.co/Lq3xAf5rMh@XavHarrison https://t.co/Nx2w81N34F@LisaDeBruine @dalejbarr https://t.co/QOrQorDD89\nSee also GLMM FAQ by @bolkerb \u0026amp; others https://t.co/s9AmINlleU pic.twitter.com/oGYC3GMw99\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) May 3, 2022 ","date":1651536000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1651536000,"objectID":"d91869d0e2c8305969c609e41c4595d4","permalink":"https://oliviergimenez.github.io/blog/2022-05-03-useful-papers-on-glmms/","publishdate":"2022-05-03T00:00:00Z","relpermalink":"/blog/2022-05-03-useful-papers-on-glmms/","section":"blog","summary":"Useful papers on GLMMs 🤩🥳@bolkerb https://t.co/HX41N05rwK@BobOHara https://t.co/c8fVvxG7ru@mattjsilk https://t.co/Lq3xAf5rMh@XavHarrison https://t.co/Nx2w81N34F@LisaDeBruine @dalejbarr https://t.co/QOrQorDD89\nSee also GLMM FAQ by @bolkerb \u0026amp; others https://t.co/s9AmINlleU pic.twitter.com/oGYC3GMw99\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) May 3, 2022 ","tags":["glmm"],"title":"Useful papers on GLMMs","type":"blog"},{"authors":null,"categories":null,"content":"Here I illustrate how to fit GLMMs with the R package brms, and compare to Jags and lme4.\nMotivation I regularly give a course on Bayesian statistics with R for non-specialists. To illustrate the course, we analyse data with generalized linear, often mixed, models or GLMMs.\nSo far, I\u0026rsquo;ve been using Jags to fit these models. This requires some programming skills, like e.g. coding a loop, to be able to write down the model likelihood. Although students learn a lot from going through that process, it can be daunting.\nThis year, I thought I\u0026rsquo;d show them the R package brms developed by Paul-Christian Bürkner. In brief, brms allows fitting GLMMs (but not only) in a lme4-like syntax within the Bayesian framework and MCMC methods with Stan.\nI\u0026rsquo;m not a Stan user, but it doesn\u0026rsquo;t matter. The vignettes were more than enough to get me started. I also recommend the list of blog posts about brms.\nFirst things first, we load the packages we will need:\nlibrary(brms) library(posterior) # tools for working with posterior and prior distributions library(R2jags) # run Jags from within R library(lme4) # fit GLMM in frequentist framework Beta-Binomial model The first example is about a survival experiment. We captured, marked and released 57 individuals (total) out of which 19 made it through the year (alive).\ndat \u0026lt;- data.frame(alive = 19, total = 57) dat An obvious estimate of the probability of a binomial is the proportion of cases, $19/57 = 0.3333333$ here, but let\u0026rsquo;s use R built-in functions for the sake of illustration.\nMaximum-likelihood estimation To get an estimate of survival, we fit a logistic regression using the glm() function. The data are grouped (or aggregated), so we need to specify the number of alive and dead individuals as a two-column matrix on the left hand side of the model formula:\nmle \u0026lt;- glm(cbind(alive, total - alive) ~ 1, family = binomial(\u0026quot;logit\u0026quot;), # family = binomial(\u0026quot;identity\u0026quot;) would be more straightforward data = dat) On the logit scale, survival is estimated at:\ncoef(mle) # logit scale ## (Intercept) ## -0.6931472 After back-transformation using the reciprocal logit, we obtain:\nplogis(coef(mle)) ## (Intercept) ## 0.3333333 So far, so good.\nBayesian analysis with Jags In Jags, you can fit this model with:\n# model betabin \u0026lt;- function(){ alive ~ dbin(survival, total) # binomial likelihood logit(survival) \u0026lt;- beta # logit link beta ~ dnorm(0, 1/1.5) # prior } # data datax \u0026lt;- list(total = 57, alive = 19) # initial values inits \u0026lt;- function() list(beta = rnorm(1, 0, sd = 1.5)) # parameter to monitor params \u0026lt;- c(\u0026quot;survival\u0026quot;) # run jags bayes.jags \u0026lt;- jags(data = datax, inits = inits, parameters.to.save = params, model.file = betabin, n.chains = 2, n.iter = 5000, n.burnin = 1000, n.thin = 1) ## Compiling model graph ## Resolving undeclared variables ## Allocating nodes ## Graph information: ## Observed stochastic nodes: 1 ## Unobserved stochastic nodes: 1 ## Total graph size: 8 ## ## Initializing model # display results bayes.jags ## Inference for Bugs model at \u0026quot;/var/folders/r7/j0wqj1k95vz8w44sdxzm986c0000gn/T//RtmpP7MZ1A/modela81d72dd23e2.txt\u0026quot;, fit using jags, ## 2 chains, each with 5000 iterations (first 1000 discarded) ## n.sims = 8000 iterations saved ## mu.vect sd.vect 2.5% 25% 50% 75% 97.5% Rhat n.eff ## survival 0.342 0.061 0.228 0.300 0.341 0.382 0.466 1.001 2600 ## deviance 5.354 1.390 4.388 4.486 4.819 5.662 9.333 1.001 8000 ## ## For each parameter, n.eff is a crude measure of effective sample size, ## and Rhat is the potential scale reduction factor (at convergence, Rhat=1). ## ## DIC info (using the rule, pD = var(deviance)/2) ## pD = 1.0 and DIC = 6.3 ## DIC is an estimate of expected predictive error (lower deviance is better). Bayesian analysis with brms In brms, you write:\nbayes.brms \u0026lt;- brm(alive | trials(total) ~ 1, family = binomial(\u0026quot;logit\u0026quot;), # binomial(\u0026quot;identity\u0026quot;) would be more straightforward data = dat, chains = 2, # nb of chains iter = 5000, # nb of iterations, including burnin warmup = 1000, # burnin thin = 1) # thinning ## Running /Library/Frameworks/R.framework/Resources/bin/R CMD SHLIB foo.c ## clang -mmacosx-version-min=10.13 -I\u0026quot;/Library/Frameworks/R.framework/Resources/include\u0026quot; -DNDEBUG -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/Rcpp/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/unsupported\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/BH/include\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/src/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppParallel/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/rstan/include\u0026quot; -DEIGEN_NO_DEBUG -DBOOST_DISABLE_ASSERTS -DBOOST_PENDING_INTEGER_LOG2_HPP -DSTAN_THREADS -DBOOST_NO_AUTO_PTR -include '/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp' -D_REENTRANT -DRCPP_PARALLEL_USE_TBB=1 -I/usr/local/include -fPIC -Wall -g -O2 -c foo.c -o foo.o ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:88: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:1: error: unknown type name 'namespace' ## namespace Eigen { ## ^ ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:16: error: expected ';' after top level declarator ## namespace Eigen { ## ^ ## ; ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:96:10: fatal error: 'complex' file not found ## #include \u0026lt;complex\u0026gt; ## ^~~~~~~~~ ## 3 errors generated. ## make: *** [foo.o] Error 1 ## ## SAMPLING FOR MODEL '782790eabbd10296f779d85fa03cf1e5' NOW (CHAIN 1). ## Chain 1: ## Chain 1: Gradient evaluation took 2e-05 seconds ## Chain 1: 1000 transitions using 10 leapfrog steps per transition would take 0.2 seconds. ## Chain 1: Adjust your expectations accordingly! ## Chain 1: ## Chain 1: ## Chain 1: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 1: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 1: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 1: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 1: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 1: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 1: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 1: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 1: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 1: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 1: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 1: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 1: ## Chain 1: Elapsed Time: 0.010358 seconds (Warm-up) ## Chain 1: 0.046464 seconds (Sampling) ## Chain 1: 0.056822 seconds (Total) ## Chain 1: ## ## SAMPLING FOR MODEL '782790eabbd10296f779d85fa03cf1e5' NOW (CHAIN 2). ## Chain 2: ## Chain 2: Gradient evaluation took 1.3e-05 seconds ## Chain 2: 1000 transitions using 10 leapfrog steps per transition would take 0.13 seconds. ## Chain 2: Adjust your expectations accordingly! ## Chain 2: ## Chain 2: ## Chain 2: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 2: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 2: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 2: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 2: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 2: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 2: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 2: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 2: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 2: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 2: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 2: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 2: ## Chain 2: Elapsed Time: 0.009763 seconds (Warm-up) ## Chain 2: 0.035949 seconds (Sampling) ## Chain 2: 0.045712 seconds (Total) ## Chain 2: You can display the results:\nbayes.brms ## Family: binomial ## Links: mu = logit ## Formula: alive | trials(total) ~ 1 ## Data: dat (Number of observations: 1) ## Draws: 2 chains, each with iter = 5000; warmup = 1000; thin = 1; ## total post-warmup draws = 8000 ## ## Population-Level Effects: ## Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS ## Intercept -0.69 0.28 -1.27 -0.14 1.00 3213 3572 ## ## Draws were sampled using sampling(NUTS). For each parameter, Bulk_ESS ## and Tail_ESS are effective sample size measures, and Rhat is the potential ## scale reduction factor on split chains (at convergence, Rhat = 1). And visualize the posterior density and trace of survival (on the logit scale):\nplot(bayes.brms) To get survival on the $[0,1]$ scale, we extract the MCMC values, then apply the reciprocal logit function to each of these values, and summarize its posterior distribution:\ndraws_fit \u0026lt;- as_draws_matrix(bayes.brms) draws_fit ## # A draws_matrix: 4000 iterations, 2 chains, and 2 variables ## variable ## draw b_Intercept lp__ ## 1 -0.70 -4.2 ## 2 -0.67 -4.2 ## 3 -0.86 -4.4 ## 4 -0.80 -4.3 ## 5 -1.32 -6.6 ## 6 -0.57 -4.2 ## 7 -0.51 -4.4 ## 8 -0.56 -4.3 ## 9 -0.69 -4.2 ## 10 -0.41 -4.6 ## # ... with 7990 more draws summarize_draws(plogis(draws_fit[,1])) What is the prior used by default?\nprior_summary(bayes.brms) What if I want to use the same prior as in Jags instead?\nnlprior \u0026lt;- prior(normal(0, 1.5), class = \u0026quot;Intercept\u0026quot;) # new prior bayes.brms \u0026lt;- brm(alive | trials(total) ~ 1, family = binomial(\u0026quot;logit\u0026quot;), data = dat, prior = nlprior, # set prior by hand chains = 2, # nb of chains iter = 5000, # nb of iterations, including burnin warmup = 1000, # burnin thin = 1) ## Running /Library/Frameworks/R.framework/Resources/bin/R CMD SHLIB foo.c ## clang -mmacosx-version-min=10.13 -I\u0026quot;/Library/Frameworks/R.framework/Resources/include\u0026quot; -DNDEBUG -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/Rcpp/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/unsupported\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/BH/include\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/src/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppParallel/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/rstan/include\u0026quot; -DEIGEN_NO_DEBUG -DBOOST_DISABLE_ASSERTS -DBOOST_PENDING_INTEGER_LOG2_HPP -DSTAN_THREADS -DBOOST_NO_AUTO_PTR -include '/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp' -D_REENTRANT -DRCPP_PARALLEL_USE_TBB=1 -I/usr/local/include -fPIC -Wall -g -O2 -c foo.c -o foo.o ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:88: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:1: error: unknown type name 'namespace' ## namespace Eigen { ## ^ ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:16: error: expected ';' after top level declarator ## namespace Eigen { ## ^ ## ; ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:96:10: fatal error: 'complex' file not found ## #include \u0026lt;complex\u0026gt; ## ^~~~~~~~~ ## 3 errors generated. ## make: *** [foo.o] Error 1 ## ## SAMPLING FOR MODEL '59dcc499a2b4c8a1b2c8f4e80e70fcce' NOW (CHAIN 1). ## Chain 1: ## Chain 1: Gradient evaluation took 1.8e-05 seconds ## Chain 1: 1000 transitions using 10 leapfrog steps per transition would take 0.18 seconds. ## Chain 1: Adjust your expectations accordingly! ## Chain 1: ## Chain 1: ## Chain 1: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 1: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 1: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 1: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 1: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 1: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 1: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 1: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 1: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 1: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 1: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 1: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 1: ## Chain 1: Elapsed Time: 0.00875 seconds (Warm-up) ## Chain 1: 0.03345 seconds (Sampling) ## Chain 1: 0.0422 seconds (Total) ## Chain 1: ## ## SAMPLING FOR MODEL '59dcc499a2b4c8a1b2c8f4e80e70fcce' NOW (CHAIN 2). ## Chain 2: ## Chain 2: Gradient evaluation took 5e-06 seconds ## Chain 2: 1000 transitions using 10 leapfrog steps per transition would take 0.05 seconds. ## Chain 2: Adjust your expectations accordingly! ## Chain 2: ## Chain 2: ## Chain 2: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 2: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 2: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 2: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 2: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 2: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 2: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 2: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 2: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 2: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 2: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 2: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 2: ## Chain 2: Elapsed Time: 0.009788 seconds (Warm-up) ## Chain 2: 0.036797 seconds (Sampling) ## Chain 2: 0.046585 seconds (Total) ## Chain 2: Double-check the prior that was used:\nprior_summary(bayes.brms) Logistic regression with covariates In this example, we ask whether annual variation in white stork breeding success can be explained by rainfall in the wintering area. Breeding success is measured by the ratio of the number of chicks over the number of pairs.\nnbchicks \u0026lt;- c(151,105,73,107,113,87,77,108,118,122,112,120,122,89,69,71,53,41,53,31,35,14,18) nbpairs \u0026lt;- c(173,164,103,113,122,112,98,121,132,136,133,137,145,117,90,80,67,54,58,39,42,23,23) rain \u0026lt;- c(67,52,88,61,32,36,72,43,92,32,86,28,57,55,66,26,28,96,48,90,86,78,87) dat \u0026lt;- data.frame(nbchicks = nbchicks, nbpairs = nbpairs, rain = (rain - mean(rain))/sd(rain)) # standardized rainfall The data are grouped.\nMaximum-likelihood estimation You can get maximum likelihood estimates with:\nmle \u0026lt;- glm(cbind(nbchicks, nbpairs - nbchicks) ~ rain, family = binomial(\u0026quot;logit\u0026quot;), data = dat) summary(mle) ## ## Call: ## glm(formula = cbind(nbchicks, nbpairs - nbchicks) ~ rain, family = binomial(\u0026quot;logit\u0026quot;), ## data = dat) ## ## Deviance Residuals: ## Min 1Q Median 3Q Max ## -5.9630 -1.3354 0.3929 1.6168 3.8800 ## ## Coefficients: ## Estimate Std. Error z value Pr(\u0026gt;|z|) ## (Intercept) 1.55436 0.05565 27.931 \u0026lt;2e-16 *** ## rain -0.14857 0.05993 -2.479 0.0132 * ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 ## ## (Dispersion parameter for binomial family taken to be 1) ## ## Null deviance: 109.25 on 22 degrees of freedom ## Residual deviance: 103.11 on 21 degrees of freedom ## AIC: 205.84 ## ## Number of Fisher Scoring iterations: 4 There is a negative effect of rainfall on breeding success:\nvisreg::visreg(mle, scale = \u0026quot;response\u0026quot;) Bayesian analysis with Jags In Jags, we can fit the same model:\n# data dat \u0026lt;- list(N = 23, # nb of years nbchicks = nbchicks, nbpairs = nbpairs, rain = (rain - mean(rain))/sd(rain)) # standardized rainfall # define model logistic \u0026lt;- function() { # likelihood for(i in 1:N){ # loop over years nbchicks[i] ~ dbin(p[i], nbpairs[i]) # binomial likelihood logit(p[i]) \u0026lt;- a + b.rain * rain[i] # prob success is linear fn of rainfall on logit scale } # priors a ~ dnorm(0,0.01) # intercept b.rain ~ dnorm(0,0.01) # slope for rainfall } # function that generates initial values inits \u0026lt;- function() list(a = rnorm(n = 1, mean = 0, sd = 5), b.rain = rnorm(n = 1, mean = 0, sd = 5)) # specify parameters that need to be estimated parameters \u0026lt;- c(\u0026quot;a\u0026quot;, \u0026quot;b.rain\u0026quot;) # run Jags bayes.jags \u0026lt;- jags(data = dat, inits = inits, parameters.to.save = parameters, model.file = logistic, n.chains = 2, n.iter = 5000, # includes burn-in! n.burnin = 1000) ## Compiling model graph ## Resolving undeclared variables ## Allocating nodes ## Graph information: ## Observed stochastic nodes: 23 ## Unobserved stochastic nodes: 2 ## Total graph size: 134 ## ## Initializing model # display results bayes.jags ## Inference for Bugs model at \u0026quot;/var/folders/r7/j0wqj1k95vz8w44sdxzm986c0000gn/T//RtmpP7MZ1A/modela81d1b01bd3f.txt\u0026quot;, fit using jags, ## 2 chains, each with 5000 iterations (first 1000 discarded), n.thin = 4 ## n.sims = 2000 iterations saved ## mu.vect sd.vect 2.5% 25% 50% 75% 97.5% Rhat n.eff ## a 1.546 0.163 1.439 1.514 1.552 1.588 1.657 1.182 2000 ## b.rain -0.152 0.121 -0.275 -0.190 -0.148 -0.106 -0.033 1.129 2000 ## deviance 213.612 330.910 201.899 202.434 203.197 204.689 209.169 1.121 2000 ## ## For each parameter, n.eff is a crude measure of effective sample size, ## and Rhat is the potential scale reduction factor (at convergence, Rhat=1). ## ## DIC info (using the rule, pD = var(deviance)/2) ## pD = 54753.9 and DIC = 54967.5 ## DIC is an estimate of expected predictive error (lower deviance is better). Bayesian analysis with brms With brms, we write:\nbayes.brms \u0026lt;- brm(nbchicks | trials(nbpairs) ~ rain, family = binomial(\u0026quot;logit\u0026quot;), data = dat, chains = 2, # nb of chains iter = 5000, # nb of iterations, including burnin warmup = 1000, # burnin thin = 1) ## Running /Library/Frameworks/R.framework/Resources/bin/R CMD SHLIB foo.c ## clang -mmacosx-version-min=10.13 -I\u0026quot;/Library/Frameworks/R.framework/Resources/include\u0026quot; -DNDEBUG -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/Rcpp/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/unsupported\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/BH/include\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/src/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppParallel/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/rstan/include\u0026quot; -DEIGEN_NO_DEBUG -DBOOST_DISABLE_ASSERTS -DBOOST_PENDING_INTEGER_LOG2_HPP -DSTAN_THREADS -DBOOST_NO_AUTO_PTR -include '/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp' -D_REENTRANT -DRCPP_PARALLEL_USE_TBB=1 -I/usr/local/include -fPIC -Wall -g -O2 -c foo.c -o foo.o ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:88: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:1: error: unknown type name 'namespace' ## namespace Eigen { ## ^ ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:16: error: expected ';' after top level declarator ## namespace Eigen { ## ^ ## ; ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:96:10: fatal error: 'complex' file not found ## #include \u0026lt;complex\u0026gt; ## ^~~~~~~~~ ## 3 errors generated. ## make: *** [foo.o] Error 1 ## ## SAMPLING FOR MODEL '70a68f0c53365768853c2791038e6352' NOW (CHAIN 1). ## Chain 1: ## Chain 1: Gradient evaluation took 4.7e-05 seconds ## Chain 1: 1000 transitions using 10 leapfrog steps per transition would take 0.47 seconds. ## Chain 1: Adjust your expectations accordingly! ## Chain 1: ## Chain 1: ## Chain 1: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 1: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 1: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 1: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 1: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 1: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 1: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 1: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 1: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 1: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 1: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 1: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 1: ## Chain 1: Elapsed Time: 0.035213 seconds (Warm-up) ## Chain 1: 0.131663 seconds (Sampling) ## Chain 1: 0.166876 seconds (Total) ## Chain 1: ## ## SAMPLING FOR MODEL '70a68f0c53365768853c2791038e6352' NOW (CHAIN 2). ## Chain 2: ## Chain 2: Gradient evaluation took 1.4e-05 seconds ## Chain 2: 1000 transitions using 10 leapfrog steps per transition would take 0.14 seconds. ## Chain 2: Adjust your expectations accordingly! ## Chain 2: ## Chain 2: ## Chain 2: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 2: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 2: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 2: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 2: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 2: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 2: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 2: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 2: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 2: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 2: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 2: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 2: ## Chain 2: Elapsed Time: 0.033715 seconds (Warm-up) ## Chain 2: 0.15264 seconds (Sampling) ## Chain 2: 0.186355 seconds (Total) ## Chain 2: Display results:\nbayes.brms ## Family: binomial ## Links: mu = logit ## Formula: nbchicks | trials(nbpairs) ~ rain ## Data: dat (Number of observations: 23) ## Draws: 2 chains, each with iter = 5000; warmup = 1000; thin = 1; ## total post-warmup draws = 8000 ## ## Population-Level Effects: ## Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS ## Intercept 1.56 0.06 1.45 1.67 1.00 6907 5043 ## rain -0.15 0.06 -0.27 -0.03 1.00 7193 5067 ## ## Draws were sampled using sampling(NUTS). For each parameter, Bulk_ESS ## and Tail_ESS are effective sample size measures, and Rhat is the potential ## scale reduction factor on split chains (at convergence, Rhat = 1). Visualize:\nplot(bayes.brms) We can also calculate the posterior probability of the rainfall effect being below zero with the hypothesis() function:\nhypothesis(bayes.brms, 'rain \u0026lt; 0') ## Hypothesis Tests for class b: ## Hypothesis Estimate Est.Error CI.Lower CI.Upper Evid.Ratio Post.Prob Star ## 1 (rain) \u0026lt; 0 -0.15 0.06 -0.25 -0.05 116.65 0.99 * ## --- ## 'CI': 90%-CI for one-sided and 95%-CI for two-sided hypotheses. ## '*': For one-sided hypotheses, the posterior probability exceeds 95%; ## for two-sided hypotheses, the value tested against lies outside the 95%-CI. ## Posterior probabilities of point hypotheses assume equal prior probabilities. These results confirm the important negative effect of rainfall on breeding success.\nLinear mixed model In this example, we have several measurements for 33 Mediterranean plant species, specifically number of seeds and biomass. We ask whether there is a linear relationship between these two variables that would hold for all species.\nRead in data, directly from the course website:\ndf \u0026lt;- read_csv2(\u0026quot;https://raw.githubusercontent.com/oliviergimenez/bayesian-stats-with-R/master/slides/dat/VMG.csv\u0026quot;) df Use relevant format for columns species Sp and biomass Vm:\ndf$Sp \u0026lt;- as_factor(df$Sp) df$Vm \u0026lt;- as.numeric(df$Vm) Define numeric vector of species, for nested indexing in Jags:\nspecies \u0026lt;- as.numeric(df$Sp) Define response variable, number of seeds:\ny \u0026lt;- log(df$NGrTotest) Standardize explanatory variable, biomass\nx \u0026lt;- (df$Vm - mean(df$Vm))/sd(df$Vm) Now build dataset:\ndat \u0026lt;- data.frame(y = y, x = x, species = species) Maximum-likelihood estimation You can get maximum likelihood estimates for a linear regression of number of seeds on biomass with a species random effect on the intercept (partial pooling):\nmle \u0026lt;- lmer(y ~ x + (1 | species), data = dat) summary(mle) ## Linear mixed model fit by REML ['lmerMod'] ## Formula: y ~ x + (1 | species) ## Data: dat ## ## REML criterion at convergence: 2642.5 ## ## Scaled residuals: ## Min 1Q Median 3Q Max ## -4.6761 -0.2704 -0.0120 0.2900 7.3163 ## ## Random effects: ## Groups Name Variance Std.Dev. ## species (Intercept) 113.146 10.637 ## Residual 9.373 3.062 ## Number of obs: 488, groups: species, 33 ## ## Fixed effects: ## Estimate Std. Error t value ## (Intercept) 14.5258 1.8574 7.821 ## x 0.4779 0.2421 1.974 ## ## Correlation of Fixed Effects: ## (Intr) ## x 0.002 Bayesian analysis with Jags In Jags, you would use:\n# partial pooling model partial_pooling \u0026lt;- function(){ # likelihood for(i in 1:n){ y[i] ~ dnorm(mu[i], tau.y) mu[i] \u0026lt;- a[species[i]] + b * x[i] } for (j in 1:nbspecies){ a[j] ~ dnorm(mu.a, tau.a) } # priors tau.y \u0026lt;- 1 / (sigma.y * sigma.y) sigma.y ~ dunif(0,100) mu.a ~ dnorm(0,0.01) tau.a \u0026lt;- 1 / (sigma.a * sigma.a) sigma.a ~ dunif(0,100) b ~ dnorm(0,0.01) } # data mydata \u0026lt;- list(y = y, x = x, n = length(y), species = species, nbspecies = length(levels(df$Sp))) # initial values inits \u0026lt;- function() list(mu.a = rnorm(1,0,5), sigma.a = runif(0,0,10), b = rnorm(1,0,5), sigma.y = runif(0,0,10)) # parameters to monitor params \u0026lt;- c(\u0026quot;mu.a\u0026quot;, \u0026quot;sigma.a\u0026quot;, \u0026quot;b\u0026quot;, \u0026quot;sigma.y\u0026quot;) # call jags to fit model bayes.jags \u0026lt;- jags(data = mydata, inits = inits, parameters.to.save = params, model.file = partial_pooling, n.chains = 2, n.iter = 5000, n.burnin = 1000, n.thin = 1) ## Compiling model graph ## Resolving undeclared variables ## Allocating nodes ## Graph information: ## Observed stochastic nodes: 488 ## Unobserved stochastic nodes: 37 ## Total graph size: 2484 ## ## Initializing model bayes.jags ## Inference for Bugs model at \u0026quot;/var/folders/r7/j0wqj1k95vz8w44sdxzm986c0000gn/T//RtmpP7MZ1A/modela81d7639dc1c.txt\u0026quot;, fit using jags, ## 2 chains, each with 5000 iterations (first 1000 discarded) ## n.sims = 8000 iterations saved ## mu.vect sd.vect 2.5% 25% 50% 75% 97.5% Rhat ## b 0.476 0.244 -0.005 0.314 0.472 0.638 0.964 1.001 ## mu.a 14.025 1.916 10.146 12.774 14.009 15.299 17.760 1.001 ## sigma.a 11.094 1.465 8.696 10.039 10.934 11.952 14.330 1.001 ## sigma.y 3.072 0.101 2.884 3.002 3.069 3.138 3.277 1.001 ## deviance 2478.206 8.704 2463.096 2472.022 2477.521 2483.650 2497.107 1.001 ## n.eff ## b 8000 ## mu.a 8000 ## sigma.a 8000 ## sigma.y 8000 ## deviance 8000 ## ## For each parameter, n.eff is a crude measure of effective sample size, ## and Rhat is the potential scale reduction factor (at convergence, Rhat=1). ## ## DIC info (using the rule, pD = var(deviance)/2) ## pD = 37.9 and DIC = 2516.1 ## DIC is an estimate of expected predictive error (lower deviance is better). Bayesian analysis with brms In brms, you write:\nbayes.brms \u0026lt;- brm(y ~ x + (1 | species), data = dat, chains = 2, # nb of chains iter = 5000, # nb of iterations, including burnin warmup = 1000, # burnin thin = 1) ## Running /Library/Frameworks/R.framework/Resources/bin/R CMD SHLIB foo.c ## clang -mmacosx-version-min=10.13 -I\u0026quot;/Library/Frameworks/R.framework/Resources/include\u0026quot; -DNDEBUG -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/Rcpp/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/unsupported\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/BH/include\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/src/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppParallel/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/rstan/include\u0026quot; -DEIGEN_NO_DEBUG -DBOOST_DISABLE_ASSERTS -DBOOST_PENDING_INTEGER_LOG2_HPP -DSTAN_THREADS -DBOOST_NO_AUTO_PTR -include '/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp' -D_REENTRANT -DRCPP_PARALLEL_USE_TBB=1 -I/usr/local/include -fPIC -Wall -g -O2 -c foo.c -o foo.o ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:88: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:1: error: unknown type name 'namespace' ## namespace Eigen { ## ^ ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:16: error: expected ';' after top level declarator ## namespace Eigen { ## ^ ## ; ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:96:10: fatal error: 'complex' file not found ## #include \u0026lt;complex\u0026gt; ## ^~~~~~~~~ ## 3 errors generated. ## make: *** [foo.o] Error 1 ## ## SAMPLING FOR MODEL 'c7dcb37f6d55803a5216866e78a4a476' NOW (CHAIN 1). ## Chain 1: ## Chain 1: Gradient evaluation took 0.000102 seconds ## Chain 1: 1000 transitions using 10 leapfrog steps per transition would take 1.02 seconds. ## Chain 1: Adjust your expectations accordingly! ## Chain 1: ## Chain 1: ## Chain 1: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 1: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 1: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 1: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 1: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 1: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 1: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 1: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 1: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 1: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 1: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 1: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 1: ## Chain 1: Elapsed Time: 0.656886 seconds (Warm-up) ## Chain 1: 3.37389 seconds (Sampling) ## Chain 1: 4.03077 seconds (Total) ## Chain 1: ## ## SAMPLING FOR MODEL 'c7dcb37f6d55803a5216866e78a4a476' NOW (CHAIN 2). ## Chain 2: ## Chain 2: Gradient evaluation took 2.7e-05 seconds ## Chain 2: 1000 transitions using 10 leapfrog steps per transition would take 0.27 seconds. ## Chain 2: Adjust your expectations accordingly! ## Chain 2: ## Chain 2: ## Chain 2: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 2: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 2: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 2: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 2: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 2: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 2: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 2: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 2: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 2: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 2: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 2: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 2: ## Chain 2: Elapsed Time: 0.593596 seconds (Warm-up) ## Chain 2: 2.34578 seconds (Sampling) ## Chain 2: 2.93937 seconds (Total) ## Chain 2: Display the results:\nbayes.brms ## Family: gaussian ## Links: mu = identity; sigma = identity ## Formula: y ~ x + (1 | species) ## Data: dat (Number of observations: 488) ## Draws: 2 chains, each with iter = 5000; warmup = 1000; thin = 1; ## total post-warmup draws = 8000 ## ## Group-Level Effects: ## ~species (Number of levels: 33) ## Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS ## sd(Intercept) 10.74 1.39 8.45 13.78 1.00 427 943 ## ## Population-Level Effects: ## Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS ## Intercept 14.13 1.90 10.31 17.86 1.01 331 580 ## x 0.47 0.25 -0.00 0.97 1.00 2317 3382 ## ## Family Specific Parameters: ## Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS ## sigma 3.07 0.10 2.88 3.28 1.00 2667 3408 ## ## Draws were sampled using sampling(NUTS). For each parameter, Bulk_ESS ## and Tail_ESS are effective sample size measures, and Rhat is the potential ## scale reduction factor on split chains (at convergence, Rhat = 1). You can extract a block of fixed effects:\nsummary(bayes.brms)$fixed And a block of random effects:\nsummary(bayes.brms)$random ## $species ## Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS ## sd(Intercept) 10.74273 1.388061 8.454543 13.776 1.001916 426.9568 943.1178 And visualize:\nplot(bayes.brms) GLMM with Poisson distribution This example is from Jason Matthiopoulos\u0026rsquo; excellent book How to be a quantitative ecologist. A survey of a coral reef uses 10 predefined linear transects covered by divers once every week. The response variable of interest is the abundance of a particular species of anemone as a function of water temperature. Counts of anemones are recorded at 20 regular line segments along the transect.\nLet\u0026rsquo;s simulate some data:\nset.seed(666) transects \u0026lt;- 10 data \u0026lt;- NULL for (tr in 1:transects){ # random effect (intercept) ref \u0026lt;- rnorm(1,0,.5) # water temperature gradient t \u0026lt;- runif(1, 18,22) + runif(1,-.2,0.2)*1:20 # Anemone gradient (expected response) ans \u0026lt;- exp(ref -14 + 1.8 * t - 0.045 * t^2) # actual counts on 20 segments of the current transect an \u0026lt;- rpois(20, ans) data \u0026lt;- rbind(data, cbind(rep(tr, 20), t, an)) } data \u0026lt;- data.frame(Transect = data[,1], Temperature = data[,2], Anemones = data[,3]) Standardize temperature:\nboo \u0026lt;- data$Temperature data$Temp \u0026lt;- (boo - mean(boo)) / sd(boo) head(data) Maximum-likelihood estimation With lme4, you write:\nfit_lme4 \u0026lt;- glmer(Anemones ~ Temp + I(Temp^2) + (1 | Transect), data = data, family = poisson) summary(fit_lme4) ## Generalized linear mixed model fit by maximum likelihood (Laplace ## Approximation) [glmerMod] ## Family: poisson ( log ) ## Formula: Anemones ~ Temp + I(Temp^2) + (1 | Transect) ## Data: data ## ## AIC BIC logLik deviance df.resid ## 1380.2 1393.4 -686.1 1372.2 196 ## ## Scaled residuals: ## Min 1Q Median 3Q Max ## -2.32356 -0.66154 -0.03391 0.63964 2.67283 ## ## Random effects: ## Groups Name Variance Std.Dev. ## Transect (Intercept) 0.1497 0.3869 ## Number of obs: 200, groups: Transect, 10 ## ## Fixed effects: ## Estimate Std. Error z value Pr(\u0026gt;|z|) ## (Intercept) 3.95436 0.12368 31.971 \u0026lt;2e-16 *** ## Temp -0.04608 0.02435 -1.892 0.0584 . ## I(Temp^2) -0.11122 0.01560 -7.130 1e-12 *** ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 ## ## Correlation of Fixed Effects: ## (Intr) Temp ## Temp 0.037 ## I(Temp^2) -0.118 -0.294 Bayesian analysis with Jags In Jags, we fit the corresponding GLMM with:\nmodel \u0026lt;- function() { for (i in 1:n){ count[i] ~ dpois(lambda[i]) log(lambda[i]) \u0026lt;- a[transect[i]] + b[1] * x[i] + b[2] * pow(x[i],2) } for (j in 1:nbtransects){ a[j] ~ dnorm (mu.a, tau.a) } mu.a ~ dnorm (0, 0.01) tau.a \u0026lt;- pow(sigma.a, -2) sigma.a ~ dunif (0, 100) b[1] ~ dnorm (0, 0.01) b[2] ~ dnorm (0, 0.01) } dat \u0026lt;- list(n = nrow(data), nbtransects = transects, x = data$Temp, count = data$Anemones, transect = data$Transect) inits \u0026lt;- function() list(a = rnorm(transects), b = rnorm(2), mu.a = rnorm(1), sigma.a = runif(1)) par \u0026lt;- c (\u0026quot;a\u0026quot;, \u0026quot;b\u0026quot;, \u0026quot;mu.a\u0026quot;, \u0026quot;sigma.a\u0026quot;) fit \u0026lt;- jags(data = dat, inits = inits, parameters.to.save = par, model.file = model, n.chains = 2, n.iter = 5000, n.burn = 1000) ## Compiling model graph ## Resolving undeclared variables ## Allocating nodes ## Graph information: ## Observed stochastic nodes: 200 ## Unobserved stochastic nodes: 14 ## Total graph size: 1622 ## ## Initializing model round(fit$BUGSoutput$summary[, -c(4,6)], 3) ## mean sd 2.5% 50% 97.5% Rhat n.eff ## a[1] 4.383 0.026 4.332 4.384 4.435 1.001 2000 ## a[2] 3.223 0.050 3.120 3.224 3.322 1.001 2000 ## a[3] 3.584 0.058 3.471 3.584 3.696 1.001 2000 ## a[4] 3.620 0.054 3.513 3.622 3.721 1.001 2000 ## a[5] 4.092 0.050 3.990 4.092 4.190 1.001 2000 ## a[6] 3.848 0.038 3.774 3.849 3.921 1.000 2000 ## a[7] 4.476 0.029 4.419 4.475 4.531 1.002 1300 ## a[8] 4.398 0.055 4.288 4.397 4.499 1.003 550 ## a[9] 3.835 0.033 3.771 3.835 3.900 1.001 1700 ## a[10] 4.076 0.029 4.018 4.076 4.132 1.001 2000 ## b[1] -0.047 0.024 -0.093 -0.047 0.000 1.001 2000 ## b[2] -0.111 0.016 -0.142 -0.111 -0.079 1.004 450 ## deviance 1324.697 4.651 1317.432 1324.185 1335.440 1.001 2000 ## mu.a 3.950 0.160 3.617 3.957 4.272 1.005 310 ## sigma.a 0.482 0.140 0.296 0.457 0.826 1.003 1000 Bayesian analysis with brms What about with brms?\nbayes.brms \u0026lt;- brm(Anemones ~ Temp + I(Temp^2) + (1 | Transect), data = data, family = poisson(\u0026quot;log\u0026quot;), chains = 2, # nb of chains iter = 5000, # nb of iterations, including burnin warmup = 1000, # burnin thin = 1) ## Running /Library/Frameworks/R.framework/Resources/bin/R CMD SHLIB foo.c ## clang -mmacosx-version-min=10.13 -I\u0026quot;/Library/Frameworks/R.framework/Resources/include\u0026quot; -DNDEBUG -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/Rcpp/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/unsupported\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/BH/include\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/src/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppParallel/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/rstan/include\u0026quot; -DEIGEN_NO_DEBUG -DBOOST_DISABLE_ASSERTS -DBOOST_PENDING_INTEGER_LOG2_HPP -DSTAN_THREADS -DBOOST_NO_AUTO_PTR -include '/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp' -D_REENTRANT -DRCPP_PARALLEL_USE_TBB=1 -I/usr/local/include -fPIC -Wall -g -O2 -c foo.c -o foo.o ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:88: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:1: error: unknown type name 'namespace' ## namespace Eigen { ## ^ ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:16: error: expected ';' after top level declarator ## namespace Eigen { ## ^ ## ; ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:96:10: fatal error: 'complex' file not found ## #include \u0026lt;complex\u0026gt; ## ^~~~~~~~~ ## 3 errors generated. ## make: *** [foo.o] Error 1 ## ## SAMPLING FOR MODEL '9326e0aa5a2dca4008c9e5b2ee4e86bf' NOW (CHAIN 1). ## Chain 1: ## Chain 1: Gradient evaluation took 4.5e-05 seconds ## Chain 1: 1000 transitions using 10 leapfrog steps per transition would take 0.45 seconds. ## Chain 1: Adjust your expectations accordingly! ## Chain 1: ## Chain 1: ## Chain 1: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 1: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 1: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 1: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 1: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 1: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 1: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 1: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 1: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 1: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 1: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 1: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 1: ## Chain 1: Elapsed Time: 0.893226 seconds (Warm-up) ## Chain 1: 4.91683 seconds (Sampling) ## Chain 1: 5.81006 seconds (Total) ## Chain 1: ## ## SAMPLING FOR MODEL '9326e0aa5a2dca4008c9e5b2ee4e86bf' NOW (CHAIN 2). ## Chain 2: ## Chain 2: Gradient evaluation took 2.4e-05 seconds ## Chain 2: 1000 transitions using 10 leapfrog steps per transition would take 0.24 seconds. ## Chain 2: Adjust your expectations accordingly! ## Chain 2: ## Chain 2: ## Chain 2: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 2: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 2: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 2: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 2: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 2: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 2: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 2: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 2: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 2: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 2: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 2: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 2: ## Chain 2: Elapsed Time: 1.21769 seconds (Warm-up) ## Chain 2: 4.45281 seconds (Sampling) ## Chain 2: 5.6705 seconds (Total) ## Chain 2: Display results:\nbayes.brms ## Family: poisson ## Links: mu = log ## Formula: Anemones ~ Temp + I(Temp^2) + (1 | Transect) ## Data: data (Number of observations: 200) ## Draws: 2 chains, each with iter = 5000; warmup = 1000; thin = 1; ## total post-warmup draws = 8000 ## ## Group-Level Effects: ## ~Transect (Number of levels: 10) ## Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS ## sd(Intercept) 0.47 0.14 0.29 0.81 1.00 1366 2089 ## ## Population-Level Effects: ## Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS ## Intercept 3.95 0.16 3.63 4.26 1.00 1359 1751 ## Temp -0.05 0.02 -0.09 0.00 1.00 4126 4490 ## ITempE2 -0.11 0.02 -0.14 -0.08 1.00 4018 4209 ## ## Draws were sampled using sampling(NUTS). For each parameter, Bulk_ESS ## and Tail_ESS are effective sample size measures, and Rhat is the potential ## scale reduction factor on split chains (at convergence, Rhat = 1). Visualize:\nplot(bayes.brms) We can assess the quality of fit of this model:\npp_check(bayes.brms, ndraws = 100, type = 'ecdf_overlay') As expected, the fit is almost perfect because we simulated data from this very model.\nWhat if we\u0026rsquo;d like to test the effect of temperature using WAIC?\nWe fit a model with no effect of temperature:\nbayes.brms2 \u0026lt;- brm(Anemones ~ 1 + (1 | Transect), data = data, family = poisson(\u0026quot;log\u0026quot;), chains = 2, # nb of chains iter = 5000, # nb of iterations, including burnin warmup = 1000, # burnin thin = 1) ## Running /Library/Frameworks/R.framework/Resources/bin/R CMD SHLIB foo.c ## clang -mmacosx-version-min=10.13 -I\u0026quot;/Library/Frameworks/R.framework/Resources/include\u0026quot; -DNDEBUG -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/Rcpp/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/unsupported\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/BH/include\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/src/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppParallel/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/rstan/include\u0026quot; -DEIGEN_NO_DEBUG -DBOOST_DISABLE_ASSERTS -DBOOST_PENDING_INTEGER_LOG2_HPP -DSTAN_THREADS -DBOOST_NO_AUTO_PTR -include '/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp' -D_REENTRANT -DRCPP_PARALLEL_USE_TBB=1 -I/usr/local/include -fPIC -Wall -g -O2 -c foo.c -o foo.o ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:88: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:1: error: unknown type name 'namespace' ## namespace Eigen { ## ^ ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:16: error: expected ';' after top level declarator ## namespace Eigen { ## ^ ## ; ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:96:10: fatal error: 'complex' file not found ## #include \u0026lt;complex\u0026gt; ## ^~~~~~~~~ ## 3 errors generated. ## make: *** [foo.o] Error 1 ## ## SAMPLING FOR MODEL 'fd9804091450c044661531e89123d1dc' NOW (CHAIN 1). ## Chain 1: ## Chain 1: Gradient evaluation took 4.4e-05 seconds ## Chain 1: 1000 transitions using 10 leapfrog steps per transition would take 0.44 seconds. ## Chain 1: Adjust your expectations accordingly! ## Chain 1: ## Chain 1: ## Chain 1: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 1: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 1: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 1: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 1: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 1: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 1: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 1: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 1: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 1: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 1: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 1: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 1: ## Chain 1: Elapsed Time: 0.704799 seconds (Warm-up) ## Chain 1: 2.58736 seconds (Sampling) ## Chain 1: 3.29216 seconds (Total) ## Chain 1: ## ## SAMPLING FOR MODEL 'fd9804091450c044661531e89123d1dc' NOW (CHAIN 2). ## Chain 2: ## Chain 2: Gradient evaluation took 1.8e-05 seconds ## Chain 2: 1000 transitions using 10 leapfrog steps per transition would take 0.18 seconds. ## Chain 2: Adjust your expectations accordingly! ## Chain 2: ## Chain 2: ## Chain 2: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 2: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 2: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 2: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 2: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 2: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 2: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 2: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 2: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 2: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 2: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 2: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 2: ## Chain 2: Elapsed Time: 0.648008 seconds (Warm-up) ## Chain 2: 2.85026 seconds (Sampling) ## Chain 2: 3.49826 seconds (Total) ## Chain 2: Then we compare both models, by ranking them with their WAIC or using ELPD:\n(waic1 \u0026lt;- waic(bayes.brms)) # waic model w/ tempterature ## ## Computed from 8000 by 200 log-likelihood matrix ## ## Estimate SE ## elpd_waic -668.2 9.8 ## p_waic 11.0 1.1 ## waic 1336.3 19.7 (waic2 \u0026lt;- waic(bayes.brms2)) # waic model wo/ tempterature ## ## Computed from 8000 by 200 log-likelihood matrix ## ## Estimate SE ## elpd_waic -702.4 13.3 ## p_waic 12.3 1.2 ## waic 1404.9 26.5 ## ## 2 (1.0%) p_waic estimates greater than 0.4. We recommend trying loo instead. loo_compare(waic1, waic2) ## elpd_diff se_diff ## bayes.brms 0.0 0.0 ## bayes.brms2 -34.3 8.9 Conclusions In all examples, you can check that i) there is little or no difference between Jags and brms numeric summaries of posterior distributions, and ii) maximum likelihood parameter estimates (glm(), lmer() and glmer() functions) and posterior means/medians are very close to each other.\nIn contrast to Jags, Nimble or Stan, you do not need to code the likelihood yourself in brms, as long as it is available in the package. You can have a look to a list of distributions with ?brms::brmsfamily, and there is also a possibility to define custom distributions with custom_family().\nNote I haven\u0026rsquo;t covered diagnostics of convergence, more here. You can also integrate brms outputs in a tidy workflow with tidybayes, and use the ggplot() magic.\nAs usual, the code is on GitHub, visit https://github.com/oliviergimenez/fit-glmm-with-brms.\nNow I guess it\u0026rsquo;s up to the students to pick their favorite Bayesian tool.\n","date":1650240000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1650240000,"objectID":"d5f7f86edeba830f6265ddb3e1c354aa","permalink":"https://oliviergimenez.github.io/blog/glmm-brms/","publishdate":"2022-04-18T00:00:00Z","relpermalink":"/blog/glmm-brms/","section":"blog","summary":"Here I illustrate how to fit GLMMs with the R package brms, and compare to Jags and lme4.\n","tags":["GLMM","bayes","R","brms","jags"],"title":"Bayesian analyses made easy: GLMMs in R package brms","type":"blog"},{"authors":[],"categories":[],"content":"8th conf of 🇫🇷 Statistical Ecology network #gdrEcoStat is about to start 🥳 We\u0026#39;re at 🏡 this year in @montpellier_ I\u0026#39;ll be tweeting today \u0026amp; tomorrow, hope it\u0026#39;ll pick your interest 😇🔇\n➡️ https://t.co/1MvnlFtSh3@INEE_CNRS @CNRS_OccitaniE @cefemontpellier @LbbeLyon @Ifremer_fr pic.twitter.com/PgeEPfx4UI\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 4, 2022 📸🥳 L\u0026#39;enregistrement vidéo des journées du #gdrEcoStat sont disponibles sur https://t.co/zptxN4JVol 😎 Jetez un coup d\u0026#39;oeil à la description de la vidéo pour accéder à une présentation par titre/orateur-trice. PDF dispo ici https://t.co/YmGutsYaQQ 😉\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 12, 2022 ","date":1649721600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1649721600,"objectID":"7ab945f6a0f716847106b5f254773ac8","permalink":"https://oliviergimenez.github.io/blog/2022-04-12-8emes-journees-du-gdr-ecostat/","publishdate":"2022-04-12T00:00:00Z","relpermalink":"/blog/2022-04-12-8emes-journees-du-gdr-ecostat/","section":"blog","summary":"8th conf of 🇫🇷 Statistical Ecology network #gdrEcoStat is about to start 🥳 We\u0026#39;re at 🏡 this year in @montpellier_ I\u0026#39;ll be tweeting today \u0026amp; tomorrow, hope it\u0026#39;ll pick your interest 😇🔇","tags":["statistical-ecology","conference"],"title":"8èmes journées du GdR EcoStat","type":"blog"},{"authors":[],"categories":[],"content":"📢🥳 With @MaudQueroue and @ValentinLauret we gave a short introduction to nimble @R_nimble nimble is a #rstats 📦 to fit models in the Bayesian framework w/ MCMC, it\u0026#39;s also a programming environment for using/coding fns/distns/samplers\n➡️ Check out https://t.co/MuUjGc4U8i pic.twitter.com/RBtu07v874\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 8, 2022 ","date":1649376000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1649376000,"objectID":"7945083d6b6a43f4ba7b6957aec6ed60","permalink":"https://oliviergimenez.github.io/blog/2022-04-08-introductory-workshop-on-nimble/","publishdate":"2022-04-08T00:00:00Z","relpermalink":"/blog/2022-04-08-introductory-workshop-on-nimble/","section":"blog","summary":"📢🥳 With @MaudQueroue and @ValentinLauret we gave a short introduction to nimble @R_nimble nimble is a #rstats 📦 to fit models in the Bayesian framework w/ MCMC, it\u0026#39;s also a programming environment for using/coding fns/distns/samplers","tags":["nimble","R","bayesian-statistics"],"title":"Introductory workshop on Nimble","type":"blog"},{"authors":[],"categories":[],"content":"A pleasure to work w/ #ComputoJournal @Computo85445972 for our paper on trade-offs bw #DeepLearning for species id \u0026amp; inference on predator-prey co-occurrence, which comes w/ a reproducible R workflow 😇https://t.co/Tgo6OJs7r0#OpenAccess #ReproducibleResearch #RStats 🧵⬇️ https://t.co/x2w73ezAcs\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 6, 2022 ","date":1649203200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1649203200,"objectID":"69b140c187d97f0d340b0806327cc72c","permalink":"https://oliviergimenez.github.io/blog/2022-04-06-new-paper-on-trade-offs-between-deep-learning-for-species-id-inference-on-predator-prey-co-occurrence/","publishdate":"2022-04-06T00:00:00Z","relpermalink":"/blog/2022-04-06-new-paper-on-trade-offs-between-deep-learning-for-species-id-inference-on-predator-prey-co-occurrence/","section":"blog","summary":"A pleasure to work w/ #ComputoJournal @Computo85445972 for our paper on trade-offs bw #DeepLearning for species id \u0026amp; inference on predator-prey co-occurrence, which comes w/ a reproducible R workflow 😇https://t.co/Tgo6OJs7r0#OpenAccess #ReproducibleResearch #RStats 🧵⬇️ https://t.","tags":["deep-learning","dynamic-occupancy","R"],"title":"New paper on trade-offs between Deep Learning for species id inference on predator-prey co-occurrence","type":"blog"},{"authors":[],"categories":[],"content":"Aaaaaaand it\u0026#39;s out 🤩🥳🐺🇫🇷 Awesome paper led by @oksanagrente \u0026quot;Wolf depredation hotspots in France: Clustering analyses adjusting for livestock availability\u0026quot; w/ @OFBiodiversite\nFree PDF ➡️ https://t.co/DilNeA8ikL@INEE_CNRS @CNRS_OccitaniE @umontpellier pic.twitter.com/ZcsbCANcw3\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 25, 2022 ","date":1645747200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1645747200,"objectID":"fdd691252bc768ca3dc3dfc5bba5be15","permalink":"https://oliviergimenez.github.io/blog/2022-02-25-new-paper-on-wolf-depredation-hotspots-in-france/","publishdate":"2022-02-25T00:00:00Z","relpermalink":"/blog/2022-02-25-new-paper-on-wolf-depredation-hotspots-in-france/","section":"blog","summary":"Aaaaaaand it\u0026#39;s out 🤩🥳🐺🇫🇷 Awesome paper led by @oksanagrente \u0026quot;Wolf depredation hotspots in France: Clustering analyses adjusting for livestock availability\u0026quot; w/ @OFBiodiversite\nFree PDF ➡️ https://t.co/DilNeA8ikL@INEE_CNRS @CNRS_OccitaniE @umontpellier pic.twitter.com/ZcsbCANcw3\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 25, 2022 ","tags":["loup","wolf"],"title":"New paper on wolf depredation hotspots in France","type":"blog"},{"authors":[],"categories":[],"content":"With advice by @fxcoudert and the help \u0026amp; patience of the @hal_fr folks, I ↗️ my #OpenAccess score 66% ➡️ 90% 🤩 https://t.co/etLjASx4fD I felt like a paleontologist digging up fossils, uploading the accepted or non-formated version of papers to public archives #OpenAccess https://t.co/y9dJhVUVE5 pic.twitter.com/ahKhNp0TgS\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 10, 2022 ","date":1644451200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1644451200,"objectID":"917a417f27827f83fbb25f460a5d8d90","permalink":"https://oliviergimenez.github.io/blog/2022-02-10-my-open-access-score/","publishdate":"2022-02-10T00:00:00Z","relpermalink":"/blog/2022-02-10-my-open-access-score/","section":"blog","summary":"With advice by @fxcoudert and the help \u0026amp; patience of the @hal_fr folks, I ↗️ my #OpenAccess score 66% ➡️ 90% 🤩 https://t.co/etLjASx4fD I felt like a paleontologist digging up fossils, uploading the accepted or non-formated version of papers to public archives #OpenAccess https://t.","tags":["open-access","open-science"],"title":"My Open Access score","type":"blog"},{"authors":[],"categories":[],"content":"Je fais une introduction aux statistiques bayésiennes avec R les 11-12 avril pour les doctorantes et doctorants @umontpellier du @CollegeDoctoUM 😇\nInscription ➡️ https://t.co/w5PFBGyHSz\nDiapos \u0026amp; vidéos ➡️ https://t.co/O5rAQTSksE#BayesianStatistics #rstats pic.twitter.com/VqIUcSLQja\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 1, 2022 ","date":1643673600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1643673600,"objectID":"b521f6143975524ce4bf96c4bf25ce1b","permalink":"https://oliviergimenez.github.io/blog/2022-02-01-introduction-aux-statistiques-bayesiennes-avec-r/","publishdate":"2022-02-01T00:00:00Z","relpermalink":"/blog/2022-02-01-introduction-aux-statistiques-bayesiennes-avec-r/","section":"blog","summary":"Je fais une introduction aux statistiques bayésiennes avec R les 11-12 avril pour les doctorantes et doctorants @umontpellier du @CollegeDoctoUM 😇\nInscription ➡️ https://t.co/w5PFBGyHSz\nDiapos \u0026amp; vidéos ➡️ https://t.co/O5rAQTSksE#BayesianStatistics #rstats pic.","tags":["statistiques-bayésiennes"],"title":"Introduction aux statistiques bayésiennes avec R","type":"blog"},{"authors":[],"categories":[],"content":"🤩😇 This year I am going to learn a lot working closely with amazing young researchers - I\u0026#39;m on ☁️9⃣ 🥳\nShort intro below 🧵 Give them a follow to stay up-to-date w/ their research (1/7)#ResearchIsLife @cefemontpellier @CNRS_OccitaniE @INEE_CNRS @umontpellier @twitthair1 pic.twitter.com/CBiL5fGvV0\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) January 15, 2022 ","date":1642204800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1642204800,"objectID":"fc4a69850abb477f243e3a12154730f1","permalink":"https://oliviergimenez.github.io/blog/2022-01-15-new-early-career-researchers-in-the-team/","publishdate":"2022-01-15T00:00:00Z","relpermalink":"/blog/2022-01-15-new-early-career-researchers-in-the-team/","section":"blog","summary":"🤩😇 This year I am going to learn a lot working closely with amazing young researchers - I\u0026#39;m on ☁️9⃣ 🥳\nShort intro below 🧵 Give them a follow to stay up-to-date w/ their research (1/7)#ResearchIsLife @cefemontpellier @CNRS_OccitaniE @INEE_CNRS @umontpellier @twitthair1 pic.","tags":["ECR","early-career-researchers"],"title":"New early career researchers in the team!","type":"blog"},{"authors":[],"categories":[],"content":"Faire une thèse est en général une expérience enrichissante pendant laquelle on apprend des tas de choses en recherche et sur soi-même. On interagit avec des gens géniaux. Mais c\u0026#39;est aussi trop souvent une expérience traumatisante. Il faut le dire. En parler. Et faire qqch 🧵1/14\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) January 8, 2022 ","date":1641600000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1641600000,"objectID":"e08fd4fff82613f8e52f87205a2fbb5d","permalink":"https://oliviergimenez.github.io/blog/2022-01-08-these-et-sante-mentale/","publishdate":"2022-01-08T00:00:00Z","relpermalink":"/blog/2022-01-08-these-et-sante-mentale/","section":"blog","summary":"Faire une thèse est en général une expérience enrichissante pendant laquelle on apprend des tas de choses en recherche et sur soi-même. On interagit avec des gens géniaux. Mais c\u0026#39;est aussi trop souvent une expérience traumatisante.","tags":["academic-life","mental-health"],"title":"Thèse et santé mentale","type":"blog"},{"authors":null,"categories":null,"content":"Here I illustrate how to train a CNN with Keras in R to predict from patients\u0026rsquo; CT scans those who will develop severe illness from Covid.\nMotivation Michael Blum tweeted about the STOIC2021 - COVID-19 AI challenge. The main goal of this challenge is to predict from the patients\u0026rsquo; CT scans who will develop severe illness from Covid.\nGiven my recent interest in machine learning, this challenge peaked my interest. Although Python is the machine learning lingua franca, it is possible to train a convolutional neural network (CNN) in R and perform (binary) image classification.\nHere, I will use an R interface to Keras that allows training neural networks. Note that the dataset shared for the challenge is big, like 280Go big, and it took me a day to download it. For the sake of illustration, I will use a similar but much lighter dataset from a Kaggle repository https://www.kaggle.com/plameneduardo/sarscov2-ctscan-dataset.\nThe code is available on GitHub as usual https://github.com/oliviergimenez/bin-image-classif.\nFirst things first, load the packages we will need.\nlibrary(tidyverse) theme_set(theme_light()) library(keras) Read in and process data We will need a function to process images, I\u0026rsquo;m stealing that one written by Spencer Palladino.\nprocess_pix \u0026lt;- function(lsf) { img \u0026lt;- lapply(lsf, image_load, grayscale = TRUE) # grayscale the image arr \u0026lt;- lapply(img, image_to_array) # turns it into an array arr_resized \u0026lt;- lapply(arr, image_array_resize, height = 100, width = 100) # resize arr_normalized \u0026lt;- normalize(arr_resized, axis = 1) #normalize to make small numbers return(arr_normalized) } Now let\u0026rsquo;s process images for patients with Covid, and do some reshaping. Idem with images for patients without Covid.\n# with covid lsf \u0026lt;- list.files(\u0026quot;dat/COVID/\u0026quot;, full.names = TRUE) covid \u0026lt;- process_pix(lsf) covid \u0026lt;- covid[,,,1] # get rid of last dim covid_reshaped \u0026lt;- array_reshape(covid, c(nrow(covid), 100*100)) # without covid lsf \u0026lt;- list.files(\u0026quot;dat/non-COVID/\u0026quot;, full.names = TRUE) ncovid \u0026lt;- process_pix(lsf) ncovid \u0026lt;- ncovid[,,,1] # get rid of last dim ncovid_reshaped \u0026lt;- array_reshape(ncovid, c(nrow(ncovid), 100*100)) We have 1252 CT scans of patients with Covid, and 1229 without.\nLet\u0026rsquo;s visualise these scans. Let\u0026rsquo;s pick a patient with Covid, and another one without.\nscancovid \u0026lt;- reshape2::melt(covid[10,,]) plotcovid \u0026lt;- scancovid %\u0026gt;% ggplot() + aes(x = Var1, y = Var2, fill = value) + geom_raster() + labs(x = NULL, y = NULL, title = \u0026quot;CT scan of a patient with covid\u0026quot;) + scale_fill_viridis_c() + theme(legend.position = \u0026quot;none\u0026quot;) scanncovid \u0026lt;- reshape2::melt(ncovid[10,,]) plotncovid \u0026lt;- scanncovid %\u0026gt;% ggplot() + aes(x = Var1, y = Var2, fill = value) + geom_raster() + labs(x = NULL, y = NULL, title = \u0026quot;CT scan of a patient without covid\u0026quot;) + scale_fill_viridis_c() + theme(legend.position = \u0026quot;none\u0026quot;) library(patchwork) plotcovid + plotncovid Put altogether and shuffle.\ndf \u0026lt;- rbind(cbind(covid_reshaped, 1), # 1 = covid cbind(ncovid_reshaped, 0)) # 0 = no covid set.seed(1234) shuffle \u0026lt;- sample(nrow(df), replace = F) df \u0026lt;- df[shuffle, ] Sounds great. We have everything we need to start training a convolutional neural network model.\nConvolutional neural network (CNN) Let\u0026rsquo;s build our training and testing datasets using a 80/20 split.\nset.seed(2022) split \u0026lt;- sample(2, nrow(df), replace = T, prob = c(0.8, 0.2)) train \u0026lt;- df[split == 1,] test \u0026lt;- df[split == 2,] train_target \u0026lt;- df[split == 1, 10001] # label in training dataset test_target \u0026lt;- df[split == 2, 10001] # label in testing dataset Now build our model. I use three layers (layer_dense() function) that I put one after the other with piping. I also use regularization (layer_dropout() function) to avoid overfitting.\nmodel \u0026lt;- keras_model_sequential() %\u0026gt;% layer_dense(units = 512, activation = \u0026quot;relu\u0026quot;) %\u0026gt;% layer_dropout(0.4) %\u0026gt;% layer_dense(units = 256, activation = \u0026quot;relu\u0026quot;) %\u0026gt;% layer_dropout(0.3) %\u0026gt;% layer_dense(units = 128, activation = \u0026quot;relu\u0026quot;) %\u0026gt;% layer_dropout(0.2) %\u0026gt;% layer_dense(units = 2, activation = 'softmax') Compile the model with defaults specific to binary classification.\nmodel %\u0026gt;% compile(optimizer = 'adam', loss = 'binary_crossentropy', metrics = c('accuracy')) We use one-hot encoding (to_categorical() function) aka dummy coding in statistics.\ntrain_label \u0026lt;- to_categorical(train_target) test_label \u0026lt;- to_categorical(test_target) Now let\u0026rsquo;s fit our model to the training dataset.\nfit_covid \u0026lt;- model %\u0026gt;% fit(x = train, y = train_label, epochs = 25, batch_size = 512, # try also 128 and 256 verbose = 2, validation_split = 0.2) A quick visualization of the performances shows that the algorithm is doing not too bad. No over/under-fitting. Accuracy and loss are fine.\nplot(fit_covid) What about the performances on the testing dataset?\nmodel %\u0026gt;% evaluate(test, test_label) ## loss accuracy ## 0.02048795 0.99795920 Let\u0026rsquo;s do some predictions on the testing dataset, and compare with ground truth.\npredictedclasses \u0026lt;- model %\u0026gt;% predict_classes(test) table(Prediction = predictedclasses, Actual = test_target) ## Actual ## Prediction 0 1 ## 0 243 0 ## 1 1 246 Pretty cool. Only one healthy patient is misclassified as being sick. Let\u0026rsquo;s save our model for further use.\nsave_model_tf(model, \u0026quot;model/covidmodel\u0026quot;) # save the model I\u0026rsquo;m happy with these results. In general however, we need to find ways to improve the performances. Check out some tips here with examples implemented in Keras with R there.\n","date":1641081600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1641081600,"objectID":"f099b0db0ad21cdc8aefcd61b2c9274d","permalink":"https://oliviergimenez.github.io/blog/image-classif/","publishdate":"2022-01-02T00:00:00Z","relpermalink":"/blog/image-classif/","section":"blog","summary":"Here I illustrate how to train a CNN with Keras in R to predict from patients\u0026rsquo; CT scans those who will develop severe illness from Covid.\n","tags":["deep learning","keras","R","rstats"],"title":"Binary image classification using Keras in R: Using CT scans to predict patients with Covid","type":"blog"},{"authors":[],"categories":[],"content":"Looking back to 2021, I don\u0026#39;t feel like emphasizing papers or projects. I guess this crazy period has reminded us of the importance of human relationships. Science is made by people, but people are not just scientists. Stay safe, and take care of others and yourself ⬇️🧵 1/11 pic.twitter.com/LUHPPnI1c0\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 21, 2021 ","date":1640044800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1640044800,"objectID":"3da43161105208b3692aa1504871327c","permalink":"https://oliviergimenez.github.io/blog/2021-12-21-looking-back-to-2021/","publishdate":"2021-12-21T00:00:00Z","relpermalink":"/blog/2021-12-21-looking-back-to-2021/","section":"blog","summary":"Looking back to 2021, I don\u0026#39;t feel like emphasizing papers or projects. I guess this crazy period has reminded us of the importance of human relationships. Science is made by people, but people are not just scientists.","tags":["retrospective"],"title":"Looking back to 2021","type":"blog"},{"authors":[],"categories":[],"content":"Under the kind influence of @FredBarraquand \u0026amp; @TiagoALOMarques, I have been thinking lately about #OpenAccess and #OpenScience. Access to science should be free. Period. My two cents. ⬇️🧵 1/11 #AcademicTwitter #AcademicChatter pic.twitter.com/n2F5p5qwvQ\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 19, 2021 ","date":1639872000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1639872000,"objectID":"499bb78830225dd1d43b6e9fabf219f3","permalink":"https://oliviergimenez.github.io/blog/2021-12-19-my-two-cents-on-open-science/","publishdate":"2021-12-19T00:00:00Z","relpermalink":"/blog/2021-12-19-my-two-cents-on-open-science/","section":"blog","summary":"Under the kind influence of @FredBarraquand \u0026amp; @TiagoALOMarques, I have been thinking lately about #OpenAccess and #OpenScience. Access to science should be free. Period. My two cents. ⬇️🧵 1/11 #AcademicTwitter #AcademicChatter pic.","tags":["open-science"],"title":"My two cents on Open Science","type":"blog"},{"authors":null,"categories":null,"content":"I have been interviewed recently by Lise Viollat who\u0026rsquo;s preparing a documentary for our national research group in statistical ecology. I have written my two cents answers to her challenging and very relevant questions. Thank you Lise for the interview !\nCan you introduce yourself? What is your job about? Your background? My work is motivated by questions on the ecology of large mammals, and large carnivores in particular (wolves, bears and lynx).\nWhat questions ? Very basic questions at first glance: How many animals? Where are they? How to explain trends in population abundance and species distribution?\nThese questions might seem simple, but the issue is that large carnivores are difficult to see in the field. We need methods to sample these populations, to manage uncertainties in counting, in detecting individuals and species, and to predict as well, where and how many. This is where statistical ecology comes in.\nI am what you might call an ecological statistician, and I work for the National Centre for Scientific Research.\nI have a rather standard background. I studied mathematics up to Bachelor\u0026rsquo;s degree. A degree that I repeated, I still had nightmares about it until recently. Then, from the Master\u0026rsquo;s degree onwards, I turned to applied mathematics and statistics. I embarked on a PhD in statistics applied to ecology, and to the dynamics of animal populations, with the estimation of survival and dispersal. As a post-doc, I spent time in England and Scotland to see what it was like elsewhere, to build my own research network, and to learn English.\nOnce recruited, I became increasingly interested in evolutionary ecology and conservation issues. I also worked more and more on issues related to large mammals, and species that interact with human activities.\nTo better understand these interactions, we need the human and social sciences. So I went back to studying sociology. I took a few courses, but I couldn\u0026rsquo;t go all the way to the Master\u0026rsquo;s degree. I am still learning on the job.\nWhat is ecology for you? I am certainly not the right person to answer this question, I never studied ecology at university. I learned from my students and colleagues, and on the job.\nI like definitions, so I\u0026rsquo;m going for it. Ecology is the study of the processes that influence the distribution and abundance of species, the interactions between these species, and the interactions between these species and their environment. Maud Quéroué\u0026rsquo;s thesis is an example of the study of these interactions between species using new statistical models.\nIt\u0026rsquo;s the science of interactions in a nutshell, and that\u0026rsquo;s what I like about this discipline. Between species, between species and their environment. By species, I also mean humans. And here we come back to this idea of mobilising ecology, human and social sciences, and the study of socio-ecosystems in which we mix ecology, politics, economics, sociology and anthropology. Gilles Maurer\u0026rsquo;s thesis is an illustration of this approach, and combines demography, genetics, economics and anthropology to study the relationship between Asian elephants and humans.\nFinally, when we talk about ecology, we must understand it in a broad sense, with the idea that we are also talking about the evolutionary processes that contribute to explaining variability in biological systems. The theses of Marlène Gamelon and Mathieu Buoro are examples of work combining ecology and evolution to better understand the dynamics of animal populations.\nHow would you define statistical ecology? And modelling? In statistical ecology, statistical is the adjective that qualifies the noun ecology. It is ecology that motivates, that asks the questions. This does not make statistics a mere toolbox, but you have to accept the statements of ecology to make relevant statistics.\nComing from mathematics and statistics, this is something that did not come naturally to me. In statistics, one often sees the data from ecology as the numerical application to show that the statistical method works in the real world (often in addition to simulated data).\nI learned the importance of motivating statistics by ecological questions from Jean-Dominique Lebreton, one of my thesis directors, who himself got it from Jean-Marie Legay who thought a lot about the place of modelling in biology, and interdisciplinarity.\nStatistical ecology therefore means applying and above all developing statistical methods and tools to answer questions in ecology. In France, the scientific community meets on these themes in the statistical ecology research group which we created in 2014.\nStatistics in general is the science of collecting, analysing and drawing conclusions from data. All three are important for ecology. How we collect data to test ecological hypotheses. How we analyse it. And how you interpret the results.\nIt is at the analysis stage that models are mostly used. A model in this context is an abstraction of reality, based on what is known, and is used to extract information using probabilities. An example of a statistical model is a simple linear regression in which we try to pass a line through a scatter plot to explain a trend.\nHow is statistical ecology useful? Why is it important to ask the questions \u0026ldquo;where are the species?\u0026rdquo;, and \u0026ldquo;how many animals are there?\u0026rdquo;? The species we study are in contact with humans and their activities. Fishing activities for bottlenose dolphins in the Mediterranean, for example, or sheep farming for wolves in France, to take another example.\nIn this context, I would say that statistical ecology and modelling serve three main purposes. To understand, to predict, and to help in decision making.\nI\u0026rsquo;ll take the example of the wolf, on which we do a lot of work with colleagues from the French Office for Biodiversity.\nFirst of all, we need to understand. We need to know the factors that explain the distribution of the species, this is information that allows us to establish the conservation status of a species. It was Julie Louvrier\u0026rsquo;s thesis that showed that the species is plastic and how it has recolonised the country.\nPredicting then. Julie also showed that it was possible to predict the short-term dispersal of the species with a mathematical model that was fitted to the data. As another example, Nina Santostasi in her thesis evaluated the effectiveness of different management scenarios for hybridisation between dogs and wolves in Italy.\nFinally, informing the management decision process. The species is managed by the French state, and it is on the basis of the numbers we produce thanks to statistical ecology that the maximum number of culls is established each year. This was Lucile Marescot\u0026rsquo;s thesis. We also need to know the effect of shooting on wolf predation of domestic livestock, a complex problem in spatial statistics. This is Oksana Grente\u0026rsquo;s recently defended thesis.\nTo sum up, there is a lot of discussion about biodiversity and ecosystem services, and therefore about the interdependence between humans and nature. A connection that is increasingly forgotten, to quote a theme dear to Alix Cosquer who is part of our team and works in environmental psychology.\nIn this context, statistical ecology is used to understand, predict and inform decision-making on the interactions between species, animals and humans, and their environment.\nWhy did you choose this career? Several reasons, in no particular order.\nIt is an incredible privilege to be paid to be a researcher, especially as a civil servant. Personally, I did not want to experience the difficult working conditions of my parents. Professionnaly, this status offers good conditions for doing our job. A permanent position allows us to work as a collective. This is an important point because research is an adventure involving several people, a team adventure; you don\u0026rsquo;t achieve anything, or not much, when you work alone (at least for me). You don\u0026rsquo;t do good research by pushing for more competition.\nOur status also allows us to take a long view, a long view that is essential for quality research, research that also serves society. Our work on the wolf, which I mentioned, began more than 10 years ago! This long time frame is being undermined by the precarization of employment which is hitting the research world hard, with more and more fixed-term contracts, and I am worried about the consequences for the younger generation of course. More competition. More administration. More excellence. Go fast. Quantity. Productivity. Fewer jobs. Fewer resources. Research and teaching are managed like a company, we have to optimise, rationalise, students are customers. It can\u0026rsquo;t work. Sorry, I\u0026rsquo;m going astray\u0026hellip; Another reason I chose this career is that I wanted to tackle applied problems, which could have concrete applications for people, for society. Or at least to contribute a little bit in that sense. To be useful, basically.\nAnother reason is that it is very (very) rewarding to accompany and support the new generation of scientists. It\u0026rsquo;s exciting, I love to walk the path with those who are embarking on a master\u0026rsquo;s degree, a thesis or a post-doc.\nFinally, this job allows you to be on the move all the time in your head. You learn all the time. You are constantly intellectually stimulated. I love that.\nDo you have to be good at math to do this job? We all have a maths hump. The sense of number, of counting. It is by training this sense that we improve, that we become better at maths. Like with a musical instrument.\nThe main obstacle is surely the language, because maths is also a language. Again like with music. You have to learn it to read it, to write it, and to understand what people who do it say to each other. With some efforts, you can really enjoy yourself.\nWhat\u0026rsquo;s important to realise is that you can use maths in ecology at any level. From the simple rule of three, which I have mastered to the use of stochastic differential equations, which I have not understood. Everyone can find what they are looking for, and do relevant research. For statistics I recommend David Spiegelhalter\u0026rsquo;s book which introduces the discipline in a very accessible way.\nWhat do you like about what you do? First of all, it\u0026rsquo;s the people I work with. My biggest motivation comes from my students. I want to work with them, to exchange, to interact. It\u0026rsquo;s all about interactions. It\u0026rsquo;s also the people in our team. I\u0026rsquo;m learning all the time. And it\u0026rsquo;s the people in the statistical ecology community in France and abroad. It\u0026rsquo;s a community that doesn\u0026rsquo;t take itself too seriously, but that does serious and exciting research to answer the big questions about the biodiversity crisis. It is also very nice to feel part of a community. By the way, we are organising the next GDR Ecologie Statistique days in Montpellier at the beginning of April 2022, don\u0026rsquo;t hesitate to join us!\nThe second thing, as I have already mentioned, is the fact that you are always on the move intellectually. You learn, you try things out, you explore, you play, you write, you read, you think, all the while feeling useful at your level in the construction of knowledge.\nThe third thing is sharing. Sharing knowledge by accompanying the students first of all, as I mentioned. And teaching too. I don\u0026rsquo;t do much as I\u0026rsquo;m already having trouble keeping my head above water with my job as a researcher. I admire my colleagues at universities who do two jobs, they are researchers and teachers! In sharing, I also include workshops that we offer over several days during which we train folks in statistical ecology methods. These workshops are free and video recorded so that as many people as possible have access to them.\nWhat are the future challenges of ecology? I have no idea! I don\u0026rsquo;t really consider myself an ecologist. The only things that come to mind are related to our research topics, of course. I think about biodiversity conservation. How do we find a balance, and a sustainable balance, between our activities and the nature around us? This means understanding better the interactions between human and non-human actors, why these interactions can lead to conflicts and how to resolve them. It also means working on complex systems, the so-called socio-ecosystems, and therefore opening up ecology even more to the human and social sciences.\nAnd the future challenges for statistics and modelling? I\u0026rsquo;m not necessarily more comfortable talking about future challenges for statistics and modelling, but we\u0026rsquo;ve thought about it a bit with colleagues from the GDR EcoStat, so I\u0026rsquo;m a bit less dry than for ecology. I see three.\nFirstly, the modelling of socio-ecosystems. There is a need to integrate different disciplines in agent-based approaches (economics, geography, ethnology). The question is how to use modelling approaches to estimate the many quantities in these complex models.\nWe have an explosion of technological means deployed at all scales of organisation. I am thinking of camera traps, environmental DNA, GPS, drones, aerial or satellite photos, etc. We are faced with a huge amount of data from which it is often difficult to extract the relevant information. The second challenge is therefore that of massive data, and the development of approaches that allow us to learn from this data and automate tasks that become impossible for humans to do. I am obviously thinking of machine learning, and deep learning in particular.\nThe third issue is predictive ecology. How can we predict ecological phenomena? This requires the manipulation of large-scale systems that couple ecology and physics, for example, or the ocean and the atmosphere, to take another example. It also requires thinking about how to propagate uncertainties throughout the analysis, right up to the prediction.\nIs there a tool, a school of thought, a discovery that has changed everything for you in your field? Yes, clearly! It is the introduction of Bayesian theory in the early 2000s. I discovered this theory as a post-doc in England and then in Scotland with my colleague Ruth King and others.\nWhat is the Bayesian theory? The idea is to update the a priori knowledge we have of a biological system with the data we accumulate on this system. This a priori knowledge can be seen as the confidence we have in a hypothesis, and the Bayesian theory allows us to update this degree of confidence thanks to the data. The Bayesian theory allows us to formalise the natural way of reasoning, to induce the chances that a hypothesis is true or false.\nThe Bayesian theory is used everywhere in science, but also in the everyday world, for example in medicine to make a diagnosis, in computer science to calibrate the anti-spam of your e-mail, in finance to manage stock portfolios, and it even allowed Alan Turing to crack the Enigma machine which allowed the Germans to communicate by coded messages.\nThe strength of the Bayesian theory also comes from the fact that it comes with very powerful algorithms to put it into practice. Very powerful in the sense that we can use increasingly complex models that are closer and closer to what we know about the species we are studying, models that we could not dream of using before.\nTo give you a concrete example, I can think of one. In ecology, we often try to model the distribution of species. This requires data on large areas to be sure that we have sampled all the habitats that are favourable or unfavourable to the species. This is difficult to do with a demanding and expensive scientific protocol. We can see that participatory science makes it possible to sample large areas at lower cost. But the data are biased because if we see a lot where there are observers, that doesn\u0026rsquo;t mean that where there is no one, the species isn\u0026rsquo;t there; it simply means that where there are people, the species is detected. Bayesian analysis makes it possible to combine these different types of monitoring to quantify distribution by combining the advantages of both types of monitoring, unbiased data from scientific protocols, and wide spatial coverage from participatory programmes. This is the thesis of Valentin Lauret.\nDon\u0026rsquo;t get me started on Bayesian theory, I teach it in our Master\u0026rsquo;s programme and to PhD students in our doctoral school and I can\u0026rsquo;t get enough of it!\nCould you give an example of your work? Recently, we\u0026rsquo;ve been working a lot on the lynx in France. The main cause of mortality for the lynx is collisions with cars. A major challenge is therefore to find ways of reducing this mortality by collision. We have developed a model that allows us to evaluate scenarios to reduce this mortality. For example, building a wildlife crossing, basically a tunnel or a bridge through which animals can cross the road without risk. We wanted this tool to be used by those involved in the lynx dossier, and in particular by the regional government. We therefore encapsulated all the modelling components in a software that can be used by non-specialists. Everything is done in R, so it is open and reproducible!\nWhy did I choose to talk about this project specifically?\nFirstly because it is a project that mixes different types of modelling, statistical and computer science, different disciplines with ecology and statistics, and sub-disciplines of ecology with landscape ecology, movement ecology, and population dynamics.\nSecondly, because it was a project in which I questioned myself a lot. We chose to show the modelling to the actors with workshops where we presented the models and where we were put in place. The stakeholders had very high expectations, both in terms of the models, their use, the outputs, the long-term maintenance of the software and the updating of the data. In the end, co-construction enabled us to be much more relevant than if we had done the same modelling work in the laboratory.\nAlso in human terms, it is a project that ended well, since the post-doct who was the link between the different partners and who did most of the work, Sarah Bauduin, was recruited by the French Office for Biodiversity.\nFinally, it is a project that opens up other research questions, with a focus on connectivity problems due to fragmentation by land transport infrastructures (car and train). This is Maëlis Kervellec\u0026rsquo;s thesis which she has just started.\nAny message for young people who might be interested in this route? You mean the old fart\u0026rsquo;s message to the younger generation? I always find it difficult to give a message, to give advice. Especially because for me it worked, but clearly I was lucky, very lucky. What worked for me will not necessarily work for others. In that sense my message and my advice are not worth much.\nI\u0026rsquo;m going to cite Ted Lasso, and I\u0026rsquo;m going to say: be interested, read, ask questions, be curious about what other people are doing, and why they are doing it. Don\u0026rsquo;t listen too much to the advice of old people, and make your own way.\nAnd have fun, don\u0026rsquo;t take yourself too seriously!\nAn anecdote to share? A rather positive consequence of confinements. It\u0026rsquo;s the story of a training course in population dynamics that we offered in 2006 with about 30 participants at that time. We\u0026rsquo;re going for a reboot next year with two colleagues and friends of the team, Aurélien Besnard and Sarah Cubaynes. One tweet and a few messages to some mailing lists later, we have 2315 subscribers after 15 days! We\u0026rsquo;re not sure how we\u0026rsquo;re going to manage all these people, but it\u0026rsquo;s very exciting! Many of the people who have signed up come from countries where access to knowledge is difficult for many reasons. So we\u0026rsquo;re doing a free, online course, and we\u0026rsquo;re recording it. That\u0026rsquo;s where I\u0026rsquo;d like to go, more knowledge sharing. It sounds a bit grandiloquent, but there is a great need for biodiversity conservation.\n📢👋 Together with @SarahCubaynes and @abesnardEPHE we will be giving a 2-day introductory workshop March 21-22 on quantitative methods for population dynamics in R #rstats Join us 🥳 It\u0026#39;s online and free of charge 😉 You just need to register https://t.co/I5sNS5kDOo Please RT 😇 pic.twitter.com/rkw8JWZBJi\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) October 11, 2021 ","date":1638748800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1638748800,"objectID":"a283cd9c5c3b79896cc376d8c98c96e8","permalink":"https://oliviergimenez.github.io/blog/interview-ecostat-english/","publishdate":"2021-12-06T00:00:00Z","relpermalink":"/blog/interview-ecostat-english/","section":"blog","summary":"I have been interviewed recently by Lise Viollat who\u0026rsquo;s preparing a documentary for our national research group in statistical ecology. I have written my two cents answers to her challenging and very relevant questions. Thank you Lise for the interview !\n","tags":["statistical ecology","interview"],"title":"Why a career in statistical ecology?","type":"blog"},{"authors":null,"categories":null,"content":"Dans le cadre d\u0026rsquo;un documentaire que réalise Lise Viollat, celle-ci m\u0026rsquo;a posé des questions sur mon parcours en recherche, et en écologie statistique. Je reproduis mes réponses ici. Merci Lise pour l\u0026rsquo;entretien !\nEst-ce que vous pouvez vous présenter? En quoi consiste votre travail, votre parcours ? Mon travail est motivé par des questions sur l’écologie des grands mammifères, et les grands carnivores en particulier (loups, ours et lynx).\nQuelles questions ? Très basiquement : combien y a-t-il d’animaux, et où sont-ils, et pourquoi ces effectifs et cette distribution.\nCes questions peuvent paraître très simples, mais le gros problème est qu’on ne peut pas facilement voir ces animaux. Il faut des méthodes pour échantillonner, pour gérer les incertitudes de comptage, de détection des individus et des espèces, et pour prédire aussi, combien et où. Et c’est là que les statistiques entrent en jeu.\nJe travaille en écologie statistique, je suis ce qu’on pourrait appeler un éco-statistien. Et je travaille au CNRS, le Centre National de la Recherche Scientifique.\nPour arriver au CNRS, j’ai eu un parcours plutôt classique. J’ai étudié les mathématiques jusqu’à la licence. Une licence que j’ai redoublée, j’en faisais encore des cauchemars jusqu’à récemment. Puis à partir du Master je me suis orienté vers les mathématiques appliquées avec les statistiques. Je me suis embarqué dans une thèse en statistique appliquée à l’écologie, et à la dynamique des populations d\u0026rsquo;animaux, avec l’estimation de la survie et de la dispersion. Après la thèse à Montpellier, j\u0026rsquo;ai fait un détour par Grenoble, puis je suis parti en Angleterre et en Ecosse pour voir comment c’était ailleurs, me faire mon propre réseau, et surtout apprendre l’anglais.\nUne fois recruté, je me suis de plus en plus intéressé aux questions d’écologie évolutive, et de conservation. J’ai aussi travaillé de plus en plus sur des questions en lien avec les grands mammifères, des espèces en interaction avec les activités humaines.\nPour mieux comprendre ces interactions, on fait appel aux sciences humaines et sociales. J’ai donc repris des études en sociologie. J\u0026rsquo;ai suivi quelques cours en licence, mais je n’ai pas pu aller au bout du Master. Je continue de me former sur le tas.\nQu\u0026rsquo;est-ce que pour vous l\u0026rsquo;écologie ? Je ne suis certainement pas la bonne personne pour répondre à la question, je n’ai jamais étudié l’écologie à l’université. J’ai appris au contact de mes étudiantes, de mes étudiants, et de mes collègues, et sur le tas.\nJ’aime bien les définitions, alors je me lance. L’écologie c’est l’étude des processus qui influencent la distribution et l’abondance des espèces, les interactions entre ces espèces, et les interactions entre ces espèces et leur environnement. La thèse de Maud Quéroué est un exemple de l\u0026rsquo;étude de ces interactions entre espèces grâce à de nouveaux modèles statistiques.\nC’est la science des interactions en bref, et c’est ça que j’aime dans cette discipline. Entre les espèces, entre les espèces et leur environnement. Dans les espèces, je mets aussi les humains. Et pour le coup on revient à cette idée de mobiliser écologie et sciences humaines et sociales, et l’étude des socio-écosystèmes dans lesquels on mélange écologie, politique, économie, sociologie et anthropologie. La thèse de Gilles Maurer est une illustration de cette démarche, et combine démographie, génétique, économie et anthropologie pour étudier les relations entre éléphant d’Asie et humains.\nPour finir, quand on parle d’écologie, il faut l’entendre au sens large, avec l’idée qu’on parle aussi des processus évolutifs qui contribuent à expliquer la variabilité dans les systèmes biologiques. Les thèses de Marlène Gamelon et Mathieu Buoro sont des exemples de travaux mêlant écologie et évolution pour mieux comprendre la dynamique des populations animales.\nComment est-ce que vous définiriez l\u0026rsquo;écologie statistique ? La modélisation ? Dans écologie statistique, l’ordre des mots est important. On ne dit pas statistique pour l’écologie. C’est l’écologie qui motive, qui pose les questions. Les statistiques n\u0026rsquo;en deviennent pas pour autant une simple boîte à outils, mais il faut accepter les énoncés de l’écologie pour faire des statistiques pertinentes.\nVenant des mathématiques et des statistiques, c’est quelque chose qui ne m’était pas naturel. En statistiques, on voit souvent les données de l’écologie comme l’application numérique pour montrer que la méthode statistique marche avec un exemple de données réelles (en complément aux données simulées souvent).\nJ’ai appris l’importance de motiver les statistiques par les questions écologiques au contact de Jean-Dominique Lebreton un de mes directeurs de thèse, qui le tenait lui-même de Jean-Marie Legay qui a beaucoup réfléchi sur la place de la modélisation en biologie, et l’interdisciplinarité.\nL’écologie statistique c’est donc appliquer et surtout développer des méthodes et des outils statistiques pour répondre aux questions de l’écologie. En France, la communauté scientifique se retrouve sur ces thèmes dans le groupement de recherche en écologie statistique que nous avons créé en 2014.\nLa statistique de manière générale, c’est la science qui permet de collecter, analyser et tirer des conclusions à partir des données. Les trois sont importants pour l’écologie. Comment on récolte des données pour tester des hypothèses écologiques. Comment on les analyse. Et comment on interprète les résultats.\nC’est à l’étape d’analyse qu’on utilise surtout des modèles. Un modèle dans ce contexte c’est une abstraction de la réalité, qui se base sur ce qu’on sait, et qu’on utilise pour extraire de l’information en utilisant les probabilités. Un exemple de modèle statistique c’est la régression linéaire simple dans laquelle on essaye de faire passer une droite dans un nuage de point pour expliquer une tendance.\nA quoi elles servent d\u0026rsquo;après vous ? Pourquoi est-il important de poser les questions \u0026ldquo;où sont les espèces?\u0026rdquo;, et \u0026ldquo;combien y a-t-il d’animaux?\u0026rdquo;. Les espèces qu’on étudie sont au contact des humains, et de leurs activités. Les activités de pêche pour le grand dauphin en Méditerranée par exemple, ou l’élévage d’ovins pour les loups en France pour prendre un autre exemple.\nDans ce contexte, je dirais que l’écologie statistique et la modélisation servent principalement à trois choses. Comprendre, prédire, et aider à la décision.\nJe prends l’exemple du loup sur lequel on travaille beaucoup avec les collègues de l’ Office Français de la Biodiversité.\nComprendre d’abord. On a besoin de connaître les facteurs qui expliquent la distribution de l’espèce, ce sont des informations qui permettent d’établir le statut de conservation par exemple, celui de l’UICN. C’était la thèse de Julie Louvrier qui a permis de montrer que l’espèce est plastique et la façon dont l’espèce a recolonisé le pays.\nPrédire ensuite. Julie a aussi montré qu’on pouvait prédire à court terme la dispersion de l’espèce avec un modèle mathématique qu’on confrontait aux données. Autre exemple, dans sa thèse, Nina Santostasi a évalué l\u0026rsquo;efficacité de différents scénarios de gestion de l\u0026rsquo;hybridation entre chiens et loups en Italie.\nEnfin, l’aide à la décision. L’espèce est gérée par l’état français, et c’est entre autres sur la base d’effectifs que nous produisons grâce à l’écologie statistique que le nombre maximal de prélèvements est établi chaque année. C’était la thèse de Lucile Marescot. On a aussi besoin de connaître l’effet des tirs mis en place sur la prédation par le loup des troupeaux domestiques, un problème complexe de statistique spatiale. C’est la thèse d’Oksana Grente qui a soutenu récemment.\nPour résumer, on parle beaucoup de biodiversité et de services rendus par les écosystèmes, et donc d’interdépendance entre les humains et la nature. Une connexion qui s’oublie de plus en plus pour reprendre un thème cher à Alix Cosquer qui travaille dans notre équipe en psychologie de l’environnement.\nDans ce contexte, l’écologie statistique sert à comprendre, à prédire et à informer la prise de décision sur les interactions entre les espèces, animaux et humains, et leur environnement.\nPourquoi avoir choisi cette carrière ? Plusieurs raisons, dans le désordre.\nC’est un privilège incroyable que d’être payé pour faire le métier de chercheur, qui plus est avec le statut de fonctionnaire. A titre personnel, je ne voulais pas connaître les conditions de travail difficiles de mes parents. A titre professionnel, ce statut offre de bonnes conditions pour faire notre métier. Un poste permanent permet de travailler en collaborant avec les collègues. C’est un point important car la recherche c’est une aventure à plusieurs, une aventure d’équipe ; on n’arrive à rien, ou à pas grand-chose, quand on travaille seul (en tous cas moi). On ne fait pas une bonne recherche en poussant à plus de compétition. Notre statut permet aussi de s’inscrire dans un temps long, ce temps long qui est indispensable à une recherche de qualité, une recherche qui sert la société aussi. Nos travaux sur le loup dont je parlais ont commencé il y a plus de 10 ans ! Ce temps long est mis à mal par la précarisation de l’emploi qui touche le monde de la recherche de plein fouet, avec de plus en plus de CDDs, et je m’inquiète des conséquences pour la jeune génération bien sûr. Plus de compétition. Plus d’administratif, plus d’excellence. Aller vite. Quantité. Productivité. Moins de postes. Moins de moyens. On gère la recherche et l’enseignement comme une entreprise, il faut optimiser, rationnaliser, les étudiants sont des clients. Ca ne peut pas marcher. Désolé je m\u0026rsquo;égare\u0026hellip;\nUne autre raison pour laquelle j\u0026rsquo;ai choisi cette carrière est que j’avais envie de m’attaquer à des problèmes appliqués, qui pourraient avoir des applications concrètes pour les gens, pour la société. Ou au moins de contribuer un peu en ce sens. Être utile en gros.\nUn autre raison est qu\u0026rsquo;il est très (très) gratifiant d’accompagner la nouvelle génération de scientifiques. C’est excitant, j’adore faire un bout de chemin avec celles et ceux qui s\u0026rsquo;embarquent dans un master, une thèse, ou un post-doc.\nEnfin, ce métier permet d’être tout le temps en mouvement dans sa tête. On apprend tout le temps. On est en permanence stimulé intellectuellement. J’adore ça.\nFaut-il être douer en math pour faire ce métier ? Ca veut dire quoi « doué en maths » en fait ? Ca fait penser à cette expression « avoir la bosse des maths ». Cette expression vient du 19ème siècle et d’une pseudo-science qui s’appelait la phrénologie. On établissait un lien entre vos capacités ou traits de caractères et les régions du cerveau, avec l’idée que ça se répercutait sur la forme du crâne. Les bons en maths avaient une bosse sur le front qui correspondait à l’endroit supposé du cerveau où se faisait les maths. C\u0026rsquo;est absurde! On a montré depuis que les mathématiques mobilisent différentes zones du cerveau.\nIl y a un bouquin très chouette de Stanislas Dehaene professeur au collège de France en psychochologie cognitive qui s’appelle justement la bosse des maths. Que dit ce bouquin? On a tous la bosse des maths. Le sens du nombre, du dénombrement. C’est en entrainant ce sens qu’on s’améliore, qu’on devient meilleur en maths. Comme avec un instrument de musique.\nLe principal obstacle est sûrement le langage, car les maths c’est aussi un langage. A nouveau comme en musique. Ca demande de l’apprendre pour la lire, l’écrire, et comprendre ce que se disent les gens qui en font. Une fois qu’on a fait nos gammes, on peut vraiment s’amuser, se faire plaisir.\nCe qui est important de réaliser, c’est qu’on peut utiliser les maths en écologie quel que soit son niveau. De la simple règle de trois que je maitrise à peu près, jusqu’à l’utilisation des équations différentielles stochastiques auxquelles je n’ai rien compris. Chacun et chacune peut y trouver son bonheur, et faire une recherche tout à fait pertinente. Pour les statistiques je conseille le livre de David Spiegelhalter qui introduit la discipline de façon très abordable.\nQu\u0026rsquo;est ce qui vous plait dans ce que vous faite ? C\u0026rsquo;est d\u0026rsquo;abord les personnes avec qui je travaille. Ma plus grosse motivation vient de mes étudiantes et de mes étudiants. J’ai envie de travailler avec eux, d’échanger, d’interagir. On en revient aux interactions. Ce sont aussi les gens de mon équipe. J’apprends tout le temps. Et ce sont les gens de la communauté en écologie statistique en France et à l’étranger. C’est une communauté qui ne se prend pas trop au sérieux, mais qui fait une recherche sérieuse et excitante pour répondre aux grandes questions qu’on se pose sur la crise de la biodiversité. C’est aussi très agréable de se sentir appartenir à une communauté. A ce propos, nous organisons les prochaines journées du GDR Ecologie Statistique à Montpellier début avril 2022, n\u0026rsquo;hésitez pas à nous rejoindre !\nLa deuxième chose, je l’ai déjà mentionné, c’est le fait d’exercer un métier dans lequel on est toujours en mouvement intellectuellement. On apprend, on essaye, on tente des choses, on explore, on joue, on écrit, on lit, on réfléchit, tout en se sentant utile à son niveau à la construction des connaissances.\nLa troisième chose c’est le partage. Le partage des connaissances via l’accompagnement des étudiantes et des étudiants d’abord, j’en ai parlé. Et l’enseignement aussi. J’en fais peu quand on compare aux collègues de l’enseignement supérieur qui forcent le respect. J’ai déjà du mal à garder la tête hors de l’eau avec mon métier de chercheur, eux en font deux, ils sont chercheurs et enseignants ! Dans le partage, je mets aussi des ateliers que nous proposons sur plusieurs jours au cours desquels on forme aux méthodes de l’écologie statistique. Ce sont des formations gratuites et filmées pour que le plus grand nombre y ait accès.\nQuels sont les futurs enjeux de l\u0026rsquo;écologie ? Je n\u0026rsquo;en sais rien du tout! Je ne me considère pas vraiment comme un écologue. Les seules choses qui me viennent à l’esprit sont en lien avec nos thématiques de recherche évidemment. Je pense à la conservation de la biodiversité. Comment trouver un équilibre, et un équilibre durable entre nos activités et la nature qui nous entoure? Ca sous-entend mieux comprendre les interactions entre les acteurs humains et non-humains, pourquoi ces interactions peuvent mener à des conflits et comment les résoudre. Ca veut dire aussi travailler sur des systèmes complexes, ces fameux socio-écosystèmes, et donc encore plus d’ouverture de l’écologie aux sciences humaines et sociales.\nDes statistiques et de la modélisation ? Je suis pas forcément plus à l’aise pour parler des enjeux en statistiques, mais on y a un peu réfléchi avec les collègues du GDR EcoStat, et je suis donc un peu moins sec que pour l’écologie. J’en vois trois.\nLa modélisation des socio-écosystèmes d’abord. Il y a besoin d’intégrer différentes disciplines dans des approches multi-agents (économie, géographie, ethnologie). On se demande comment utiliser les approches de modélisation pour estimer les quantités importantes de ces modèles complexes.\nOn a une explosion des moyens technologiques déployés à toutes les échelles d’organisation. Je pense aux pièges photos, à l\u0026rsquo;ADN environnemental, aux GPS, aux drones, aux photos aériennes ou satellites, etc. Nous sommes face à une masse conséquente de données dont il est souvent difficile d’extraire l’information pertinente. Le deuxième enjeu est donc celui des données massives, et du développement d’approches qui permettent d’apprendre de ces données, et d’automatiser des tâches qui deviennent impossibles à faire par des humains. Je pense évidemment au machine learning, et au deep learning en particulier.\nLe troisième enjeu est celui de l’écologie prédictive. Comment prédire les phénomènes écologiques? Ca demande de manipuler des systèmes de grandes dimensions qui couplent écologie et physique par exemple, ou océan et atmosphère pour prendre un autre exemple. Ca demande aussi de réfléchir à comment propager les incertitudes d’estimation et d’échantillonnage tout le long de l’analyse, jusqu’à la prédiction.\nY a-t-il un outil, un courant de pensée, une découverte qui pour vous a tout changé dans votre domaine ? Oui, clairement! C’est l’introduction de la théorie bayésienne au début des années 2000. J’ai découvert cette théorie en post-doc en Angleterre puis en Ecosse avec ma collègue Ruth King et d’autres.\nQu’est-ce que le bayésien ? L’idée est de mettre à jour la connaissance a priori qu’on a d’un système qu’on étudie grâce à l’expérimentation, avec les données qu’on accumule sur ce système. Cette connaissance a priori, on peut la voir comme la confiance qu’on a en une hypothèse, et le bayésien permet alors de mettre à jour ce degré de confiance grâce aux données. Le bayésien permet de formaliser la façon naturelle qu’on a de raisonner, d’induire les chances qu’une hypothèse soit vraie ou fausse.\nLa théorie bayésienne est utilisée un peu partout en sciences, mais aussi dans le monde de tous les jours par exemple en médecine pour faire un diagnostic, en informatique pour calibrer l’anti-spam de votre boîte électronique, en finance pour gérer des portfolios d’actions, et elle a même permis à Alan Turing de craquer la machine Enigma qui permettait aux allemands de communiquer par messages codés.\nLa force de la théorie bayésienne vient aussi du fait qu’elle vient avec des algorithmes pour la mettre en pratique qui sont très puissants. Très puissants au sens où on peut utiliser des modèles de plus en plus complexes et de plus en plus proches de ce qu’on sait des espèces qu’on étudie, des modèles qu’on ne pouvait pas rêver d’utiliser avant.\nPour vous donner du concret, je pense à un exemple. En écologie on cherche souvent à modéliser la distribution des espèces. Ca demande d’avoir des données sur des grandes surfaces pour être sûr qu’on a échantillonné tous les habitats favorables ou pas à l’espèce. C’est difficile à faire avec un protocole scientifique exigeant et coûteux. On voit que les sciences participatives permettent d’échantillonner à moindre coût sur de grandes surfaces. Mais les données sont biaisées car si on voit beaucoup là où il y a des observateurs et observatrices, ça ne veut pas dire que là où il n’y a personne, il n’y a pas l’espèce ; ça veut simplement dire que là où on a des gens, l’espèce est détectée. L’analyse bayésienne permet de combiner ces différents suivis pour quantifier la distribution en combinant les avantages des deux types de suivi, des données non-biaisées par les protocoles scientifiques, et une grande couverture spatiale par les programmes participatifs. C\u0026rsquo;est la thèse de Valentin Lauret.\nIl ne faut pas me lancer sur la théorie bayésienne, je l\u0026rsquo;enseigne en Master et pour les doctorants et doctorantes de notre école doctorale et je ne m\u0026rsquo;en lasse pas!\nPourriez vous donnez un exemple d\u0026rsquo;un de vos travaux ? C’est un travail sur le lynx en France. La première cause de mortalité du lynx sont les collisions avec les voitures. Un gros enjeu consiste donc à trouver des moyens de réduire cette mortalité par collision. On a développé un modèle qui permet d’évaluer des scénarios pour réduire cette mortalité. Par exemple construire un passage à faune, en gros un tunnel ou un pont par lequel les animaux peuvent traverser la route sans risque. On a voulu que cet outil puisse être utilisé par les acteurs du dossier lynx, et en particulier par l’état en région via les DREAL (Direction régionale de l\u0026rsquo;Environnement, de l\u0026rsquo;Aménagement et du Logement). On a donc encapsulé toutes les composantes modélisation dans un logiciel utilisable par des non-spécialistes. Tout est fait dans le logiciel R, donc le tout est ouvert et reproductible !\nPourquoi j’ai choisi ce projet ?\nD\u0026rsquo;abord parce que c\u0026rsquo;est un projet qui mélange différents types de modélisation, statistique et informatique, différentes disciplines avec écologie et statistique, et sous-disciplines de l’écologie avec écologie du paysage, écologie du déplacement, et la dynamique des populations.\nEnsuite parce qu\u0026rsquo;humainement c’est un projet au cours duquel je me suis beaucoup remis en question. On a choisi de montrer la modélisation aux acteurs avec des ateliers où on a présenté les modèles, et où on s’est fait remettre en place. Les acteurs avaient un niveau d’exigence très élevé, à la fois sur les modèles, sur leur utilisation, les sorties, le maintien à long terme du logiciel et la mise à jour des données. Au final, la co-construction nous a permis d’être beaucoup plus pertinents que si on avait fait le même travail de modélisation au laboratoire.\nToujours humainement, c’est un projet qui finit bien, puisque la post-doctorante qui a été le trait d’union entre les différents partenaires, et qui a fait l’essentiel du travail, Sarah Bauduin, a été recrutée à l’Office Français de la Biodiversité.\nEnfin, c’est un projet qui ouvre sur d’autres questions de recherche, avec un focus sur les problèmes de connectivité dus à la fragmentation par les infrastructures de transports terrestres (voiture et train). C’est la thèse de Maëlis Kervellec qu’elle vient tout juste de démarrer.\nUn message pour les jeunes qui pourrait être intéresser par cette voie ? Vous voulez dire le message du vieux con à la jeune génération? Je trouve toujours difficile de passer un message, de donner un conseil. D’autant que pour moi ça a marché, mais clairement j’ai eu de la chance, beaucoup de chance. Ce qui a marché pour moi ne marchera pas forcément pour d’autres. En ce sens mon message et mes conseils ne valent pas grand-chose.\nJe vais vous faire du Ted Lasso, et je dirais : intéressez-vous, lisez, posez des questions, soyez curieux de ce que font les autres, et pourquoi ils le font. N’écoutez pas trop les conseils des vieux, et faites votre propre chemin.\nEt rigolez, ne vous prenez pas trop au sérieux !\nUne anecdote à partager ? Une conséquence plutôt positive des confinements. C’est l’histoire d’une formation en dynamique des populations que nous avons proposée en 2006 avec une 30aine de participants à l’époque. J’ai relancé ça avec deux collègues et amis de l’équipe, Aurélien Besnard et Sarah Cubaynes. Un tweet et deux messages à des mailing lists plus tard, on en est à 2315 inscrits au bout de 15 jours ! On ne sait pas trop comment on va gérer tous ces gens, mais c’est très excitant ! Beaucoup de personnes inscrites viennent de pays où l\u0026rsquo;accès à la connaissance est difficile pour des tas de raisons. On fait donc une formation gratuite et en ligne, et on l’enregistre. C’est vers là que j’aimerais aller, plus de partage de la connaissance. Ça fait un peu grandiloquent, mais les besoins sont criants pour la conservation de la biodiversité.\n📢👋 Together with @SarahCubaynes and @abesnardEPHE we will be giving a 2-day introductory workshop March 21-22 on quantitative methods for population dynamics in R #rstats Join us 🥳 It\u0026#39;s online and free of charge 😉 You just need to register https://t.co/I5sNS5kDOo Please RT 😇 pic.twitter.com/rkw8JWZBJi\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) October 11, 2021 ","date":1638144000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1638144000,"objectID":"1c237477f79151fbe0b06632e7aaadf1","permalink":"https://oliviergimenez.github.io/blog/interview-ecostat/","publishdate":"2021-11-29T00:00:00Z","relpermalink":"/blog/interview-ecostat/","section":"blog","summary":"Dans le cadre d\u0026rsquo;un documentaire que réalise Lise Viollat, celle-ci m\u0026rsquo;a posé des questions sur mon parcours en recherche, et en écologie statistique. Je reproduis mes réponses ici. Merci Lise pour l\u0026rsquo;entretien !\n","tags":["statistical ecology","interview"],"title":"Pourquoi une carrière en écologie statistique?","type":"blog"},{"authors":null,"categories":null,"content":"I have a draft chapter on Bayes stats and MCMC at https://oliviergimenez.github.io/banana-book/crashcourse.html I\u0026rsquo;d love your feedback about what is confusing and what is missing 😇 #rstats\n","date":1632096000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1632096000,"objectID":"428b3e5a1631184039963f0cf1afb04d","permalink":"https://oliviergimenez.github.io/blog/draft-bayesmcmc/","publishdate":"2021-09-20T00:00:00Z","relpermalink":"/blog/draft-bayesmcmc/","section":"blog","summary":"I have a draft chapter on Bayes stats and MCMC at https://oliviergimenez.github.io/banana-book/crashcourse.html I\u0026rsquo;d love your feedback about what is confusing and what is missing 😇 #rstats\n","tags":["capture-recapture","Bayesian statistics","R","rstats"],"title":"Draft chapter on Bayes stats and MCMC in R","type":"blog"},{"authors":null,"categories":null,"content":"So, I\u0026rsquo;m writing a book 🤯 It\u0026rsquo;'\u0026rsquo;s called \u0026ldquo;Bayesian analysis of capture-recapture data with hidden Markov models - Theory and case studies in R\u0026rdquo;.\nSo, I\u0026#39;m writing a book 🤯 It\u0026#39;s called \u0026#39;Bayesian analysis of capture-recapture data w hidden Markov models - Theory and case studies in R\u0026#39;. Online version here https://t.co/Crgx5NX70s to be published w @CRCPress Will share chapter drafts 😇 \u0026amp; random thoughts along the way #rstats pic.twitter.com/ms6dIjvY7m\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) September 19, 2021 ","date":1632009600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1632009600,"objectID":"6df991e86f32cb717c3ca61c8d3c1342","permalink":"https://oliviergimenez.github.io/blog/banana-book/","publishdate":"2021-09-19T00:00:00Z","relpermalink":"/blog/banana-book/","section":"blog","summary":"So, I\u0026rsquo;m writing a book 🤯 It\u0026rsquo;'\u0026rsquo;s called \u0026ldquo;Bayesian analysis of capture-recapture data with hidden Markov models - Theory and case studies in R\u0026rdquo;.\n","tags":["capture-recapture","Bayesian statistics","R","rstats"],"title":"Bayesian analysis of capture-recapture data with hidden Markov models - Theory and case studies in R","type":"blog"},{"authors":null,"categories":null,"content":"I would like to familiarize myself with machine learning (ML) techniques in R. So I have been reading and learning by doing. I thought I\u0026rsquo;d share my experience for others who\u0026rsquo;d like to give it a try1.\nFirst version August 13, 2021, updated August 23, 2021 Since my first post, I’ve been reading notebooks shared by folks who ranked high in the challenge, and added two features that they used. Eventually, these new predictors did not help (I must be doing something wrong). I also explored some other ML algorithms. Last, I tuned the parameters more efficiently with a clever grid-search algorithm. All in all, I slightly improved my score, but most importantly, I now have a clean template for further use.\nMotivation All material available from GitHub at https://github.com/oliviergimenez/learning-machine-learning.\nThe two great books I’m using are:\nAn Introduction to Statistical Learning with Applications in R by Gareth James, Daniela Witten, Trevor Hastie and Robert Tibshirani\nTidy models in R by Max Kuhn and Julia Silge\nI also recommend checking out the material (codes, screencasts) shared by David Robinson and Julia Silge from whom I picked some useful tricks that I put to use below.\nTo try things, I’ve joined the Kaggle online community which gathers folks with lots of experience in ML from whom you can learn. Kaggle also hosts public datasets that can be used for playing around.\nI use the tidymodels metapackage that contains a suite of packages for modeling and machine learning using tidyverse principles. Check out all possibilities here, and parsnip models in particular there.\nLet’s start with the famous Titanic dataset. We need to predict if a passenger survived the sinking of the Titanic (1) or not (0). A dataset is provided for training our models (train.csv). Another dataset is provided (test.csv) for which we do not know the answer. We will predict survival for each passenger, submit our answer to Kaggle and see how well we did compared to other folks. The metric for comparison is the percentage of passengers we correctly predict – aka as accuracy.\nFirst things first, let’s load some packages to get us started.\nlibrary(tidymodels) # metapackage for ML library(tidyverse) # metapackage for data manipulation and visulaisation library(stacks) # stack ML models for better perfomance theme_set(theme_light()) doParallel::registerDoParallel(cores = 4) # parallel computations Data Read in training data.\nrawdata \u0026lt;- read_csv(\u0026quot;dat/titanic/train.csv\u0026quot;) glimpse(rawdata) ## Rows: 891 ## Columns: 12 ## $ PassengerId \u0026lt;dbl\u0026gt; 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20… ## $ Survived \u0026lt;dbl\u0026gt; 0, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, … ## $ Pclass \u0026lt;dbl\u0026gt; 3, 1, 3, 1, 3, 3, 1, 3, 3, 2, 3, 1, 3, 3, 3, 2, 3, 2, 3, 3, 2, 2, 3, … ## $ Name \u0026lt;chr\u0026gt; \u0026quot;Braund, Mr. Owen Harris\u0026quot;, \u0026quot;Cumings, Mrs. John Bradley (Florence Brig… ## $ Sex \u0026lt;chr\u0026gt; \u0026quot;male\u0026quot;, \u0026quot;female\u0026quot;, \u0026quot;female\u0026quot;, \u0026quot;female\u0026quot;, \u0026quot;male\u0026quot;, \u0026quot;male\u0026quot;, \u0026quot;male\u0026quot;, \u0026quot;male\u0026quot;,… ## $ Age \u0026lt;dbl\u0026gt; 22, 38, 26, 35, 35, NA, 54, 2, 27, 14, 4, 58, 20, 39, 14, 55, 2, NA, … ## $ SibSp \u0026lt;dbl\u0026gt; 1, 1, 0, 1, 0, 0, 0, 3, 0, 1, 1, 0, 0, 1, 0, 0, 4, 0, 1, 0, 0, 0, 0, … ## $ Parch \u0026lt;dbl\u0026gt; 0, 0, 0, 0, 0, 0, 0, 1, 2, 0, 1, 0, 0, 5, 0, 0, 1, 0, 0, 0, 0, 0, 0, … ## $ Ticket \u0026lt;chr\u0026gt; \u0026quot;A/5 21171\u0026quot;, \u0026quot;PC 17599\u0026quot;, \u0026quot;STON/O2. 3101282\u0026quot;, \u0026quot;113803\u0026quot;, \u0026quot;373450\u0026quot;, \u0026quot;330… ## $ Fare \u0026lt;dbl\u0026gt; 7.2500, 71.2833, 7.9250, 53.1000, 8.0500, 8.4583, 51.8625, 21.0750, 1… ## $ Cabin \u0026lt;chr\u0026gt; NA, \u0026quot;C85\u0026quot;, NA, \u0026quot;C123\u0026quot;, NA, NA, \u0026quot;E46\u0026quot;, NA, NA, NA, \u0026quot;G6\u0026quot;, \u0026quot;C103\u0026quot;, NA, N… ## $ Embarked \u0026lt;chr\u0026gt; \u0026quot;S\u0026quot;, \u0026quot;C\u0026quot;, \u0026quot;S\u0026quot;, \u0026quot;S\u0026quot;, \u0026quot;S\u0026quot;, \u0026quot;Q\u0026quot;, \u0026quot;S\u0026quot;, \u0026quot;S\u0026quot;, \u0026quot;S\u0026quot;, \u0026quot;C\u0026quot;, \u0026quot;S\u0026quot;, \u0026quot;S\u0026quot;, \u0026quot;S\u0026quot;, \u0026quot;S\u0026quot;,… naniar::miss_var_summary(rawdata) ## # A tibble: 12 × 3 ## variable n_miss pct_miss ## \u0026lt;chr\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; ## 1 Cabin 687 77.1 ## 2 Age 177 19.9 ## 3 Embarked 2 0.224 ## 4 PassengerId 0 0 ## 5 Survived 0 0 ## 6 Pclass 0 0 ## 7 Name 0 0 ## 8 Sex 0 0 ## 9 SibSp 0 0 ## 10 Parch 0 0 ## 11 Ticket 0 0 ## 12 Fare 0 0 After some data exploration (not shown), I decided to take care of missing values, gather the two family variables in a single variable, and create a variable title.\n# Get most frequent port of embarkation uniqx \u0026lt;- unique(na.omit(rawdata$Embarked)) mode_embarked \u0026lt;- as.character(fct_drop(uniqx[which.max(tabulate(match(rawdata$Embarked, uniqx)))])) # Build function for data cleaning and handling NAs process_data \u0026lt;- function(tbl){ tbl %\u0026gt;% mutate(class = case_when(Pclass == 1 ~ \u0026quot;first\u0026quot;, Pclass == 2 ~ \u0026quot;second\u0026quot;, Pclass == 3 ~ \u0026quot;third\u0026quot;), class = as_factor(class), gender = factor(Sex), fare = Fare, age = Age, ticket = Ticket, alone = if_else(SibSp + Parch == 0, \u0026quot;yes\u0026quot;, \u0026quot;no\u0026quot;), # alone variable alone = as_factor(alone), port = factor(Embarked), # rename embarked as port title = str_extract(Name, \u0026quot;[A-Za-z]+\\\\.\u0026quot;), # title variable title = fct_lump(title, 4)) %\u0026gt;% # keep only most frequent levels of title mutate(port = ifelse(is.na(port), mode_embarked, port), # deal w/ NAs in port (replace by mode) port = as_factor(port)) %\u0026gt;% group_by(title) %\u0026gt;% mutate(median_age_title = median(age, na.rm = T)) %\u0026gt;% ungroup() %\u0026gt;% mutate(age = if_else(is.na(age), median_age_title, age)) %\u0026gt;% # deal w/ NAs in age (replace by median in title) mutate(ticketfreq = ave(1:nrow(.), FUN = length), fareadjusted = fare / ticketfreq) %\u0026gt;% mutate(familyage = SibSp + Parch + 1 + age/70) } # Process the data dataset \u0026lt;- rawdata %\u0026gt;% process_data() %\u0026gt;% mutate(survived = as_factor(if_else(Survived == 1, \u0026quot;yes\u0026quot;, \u0026quot;no\u0026quot;))) %\u0026gt;% mutate(survived = relevel(survived, ref = \u0026quot;yes\u0026quot;)) %\u0026gt;% # first event is survived = yes select(survived, class, gender, age, alone, port, title, fareadjusted, familyage) # Have a look again glimpse(dataset) ## Rows: 891 ## Columns: 9 ## $ survived \u0026lt;fct\u0026gt; no, yes, yes, yes, no, no, no, no, yes, yes, yes, yes, no, no, no, y… ## $ class \u0026lt;fct\u0026gt; third, first, third, first, third, third, first, third, third, secon… ## $ gender \u0026lt;fct\u0026gt; male, female, female, female, male, male, male, male, female, female… ## $ age \u0026lt;dbl\u0026gt; 22, 38, 26, 35, 35, 30, 54, 2, 27, 14, 4, 58, 20, 39, 14, 55, 2, 30,… ## $ alone \u0026lt;fct\u0026gt; no, no, yes, no, yes, yes, yes, no, no, no, no, yes, yes, no, yes, y… ## $ port \u0026lt;fct\u0026gt; 3, 1, 3, 3, 3, 2, 3, 3, 3, 1, 3, 3, 3, 3, 3, 3, 2, 3, 3, 1, 3, 3, 2,… ## $ title \u0026lt;fct\u0026gt; Mr., Mrs., Miss., Mrs., Mr., Mr., Mr., Master., Mrs., Mrs., Miss., M… ## $ fareadjusted \u0026lt;dbl\u0026gt; 0.008136925, 0.080003704, 0.008894501, 0.059595960, 0.009034792, 0.0… ## $ familyage \u0026lt;dbl\u0026gt; 2.314286, 2.542857, 1.371429, 2.500000, 1.500000, 1.428571, 1.771429… naniar::miss_var_summary(dataset) ## # A tibble: 9 × 3 ## variable n_miss pct_miss ## \u0026lt;chr\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; ## 1 survived 0 0 ## 2 class 0 0 ## 3 gender 0 0 ## 4 age 0 0 ## 5 alone 0 0 ## 6 port 0 0 ## 7 title 0 0 ## 8 fareadjusted 0 0 ## 9 familyage 0 0 Let’s apply the same treatment to the test dataset.\nrawdata \u0026lt;- read_csv(\u0026quot;dat/titanic/test.csv\u0026quot;) holdout \u0026lt;- rawdata %\u0026gt;% process_data() %\u0026gt;% select(PassengerId, class, gender, age, alone, port, title, fareadjusted, familyage) glimpse(holdout) ## Rows: 418 ## Columns: 9 ## $ PassengerId \u0026lt;dbl\u0026gt; 892, 893, 894, 895, 896, 897, 898, 899, 900, 901, 902, 903, 904, 905… ## $ class \u0026lt;fct\u0026gt; third, third, second, third, third, third, third, second, third, thi… ## $ gender \u0026lt;fct\u0026gt; male, female, male, male, female, male, female, male, female, male, … ## $ age \u0026lt;dbl\u0026gt; 34.5, 47.0, 62.0, 27.0, 22.0, 14.0, 30.0, 26.0, 18.0, 21.0, 28.5, 46… ## $ alone \u0026lt;fct\u0026gt; yes, no, yes, yes, no, yes, yes, no, yes, no, yes, yes, no, no, no, … ## $ port \u0026lt;fct\u0026gt; 2, 3, 2, 3, 3, 3, 2, 3, 1, 3, 3, 3, 3, 3, 3, 1, 2, 1, 3, 1, 1, 3, 3,… ## $ title \u0026lt;fct\u0026gt; Mr., Mrs., Mr., Mr., Mrs., Mr., Miss., Mr., Mrs., Mr., Mr., Mr., Mrs… ## $ fareadjusted \u0026lt;dbl\u0026gt; 0.018730144, 0.016746411, 0.023175837, 0.020723684, 0.029395933, 0.0… ## $ familyage \u0026lt;dbl\u0026gt; 1.492857, 2.671429, 1.885714, 1.385714, 3.314286, 1.200000, 1.428571… naniar::miss_var_summary(holdout) ## # A tibble: 9 × 3 ## variable n_miss pct_miss ## \u0026lt;chr\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; ## 1 fareadjusted 1 0.239 ## 2 PassengerId 0 0 ## 3 class 0 0 ## 4 gender 0 0 ## 5 age 0 0 ## 6 alone 0 0 ## 7 port 0 0 ## 8 title 0 0 ## 9 familyage 0 0 Exploratory data analysis skimr::skim(dataset) Data summary Name dataset Number of rows 891 Number of columns 9 _______________________ Column type frequency: factor 6 numeric 3 ________________________ Group variables None Data summary\nVariable type: factor\nskim_variable n_missing complete_rate ordered n_unique top_counts survived 0 1 FALSE 2 no: 549, yes: 342 class 0 1 FALSE 3 thi: 491, fir: 216, sec: 184 gender 0 1 FALSE 2 mal: 577, fem: 314 alone 0 1 FALSE 2 yes: 537, no: 354 port 0 1 FALSE 4 3: 644, 1: 168, 2: 77, S: 2 title 0 1 FALSE 5 Mr.: 517, Mis: 182, Mrs: 125, Mas: 40 Variable type: numeric\nskim_variable n_missing complete_rate mean sd p0 p25 p50 p75 p100 hist age 0 1 29.39 13.26 0.42 21.00 30.00 35.00 80.00 ▂▇▃▁▁ fareadjusted 0 1 0.04 0.06 0.00 0.01 0.02 0.03 0.58 ▇▁▁▁▁ familyage 0 1 2.32 1.57 1.07 1.41 1.57 2.62 11.43 ▇▁▁▁▁ Let’s explore the data.\ndataset %\u0026gt;% count(survived) ## # A tibble: 2 × 2 ## survived n ## \u0026lt;fct\u0026gt; \u0026lt;int\u0026gt; ## 1 yes 342 ## 2 no 549 dataset %\u0026gt;% group_by(gender) %\u0026gt;% summarize(n = n(), n_surv = sum(survived == \u0026quot;yes\u0026quot;), pct_surv = n_surv / n) ## # A tibble: 2 × 4 ## gender n n_surv pct_surv ## \u0026lt;fct\u0026gt; \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; ## 1 female 314 233 0.742 ## 2 male 577 109 0.189 dataset %\u0026gt;% group_by(title) %\u0026gt;% summarize(n = n(), n_surv = sum(survived == \u0026quot;yes\u0026quot;), pct_surv = n_surv / n) %\u0026gt;% arrange(desc(pct_surv)) ## # A tibble: 5 × 4 ## title n n_surv pct_surv ## \u0026lt;fct\u0026gt; \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; ## 1 Mrs. 125 99 0.792 ## 2 Miss. 182 127 0.698 ## 3 Master. 40 23 0.575 ## 4 Other 27 12 0.444 ## 5 Mr. 517 81 0.157 dataset %\u0026gt;% group_by(class, gender) %\u0026gt;% summarize(n = n(), n_surv = sum(survived == \u0026quot;yes\u0026quot;), pct_surv = n_surv / n) %\u0026gt;% arrange(desc(pct_surv)) ## # A tibble: 6 × 5 ## # Groups: class [3] ## class gender n n_surv pct_surv ## \u0026lt;fct\u0026gt; \u0026lt;fct\u0026gt; \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; ## 1 first female 94 91 0.968 ## 2 second female 76 70 0.921 ## 3 third female 144 72 0.5 ## 4 first male 122 45 0.369 ## 5 second male 108 17 0.157 ## 6 third male 347 47 0.135 Some informative graphs.\ndataset %\u0026gt;% group_by(class, gender) %\u0026gt;% summarize(n = n(), n_surv = sum(survived == \u0026quot;yes\u0026quot;), pct_surv = n_surv / n) %\u0026gt;% mutate(class = fct_reorder(class, pct_surv)) %\u0026gt;% ggplot(aes(pct_surv, class, fill = class, color = class)) + geom_col(position = position_dodge()) + scale_x_continuous(labels = percent) + labs(x = \u0026quot;% in category that survived\u0026quot;, fill = NULL, color = NULL, y = NULL) + facet_wrap(~gender) dataset %\u0026gt;% mutate(age = cut(age, breaks = c(0, 20, 40, 60, 80))) %\u0026gt;% group_by(age, gender) %\u0026gt;% summarize(n = n(), n_surv = sum(survived == \u0026quot;yes\u0026quot;), pct_surv = n_surv / n) %\u0026gt;% mutate(age = fct_reorder(age, pct_surv)) %\u0026gt;% ggplot(aes(pct_surv, age, fill = age, color = age)) + geom_col(position = position_dodge()) + scale_x_continuous(labels = percent) + labs(x = \u0026quot;% in category that survived\u0026quot;, fill = NULL, color = NULL, y = NULL) + facet_wrap(~gender) dataset %\u0026gt;% ggplot(aes(fareadjusted, group = survived, color = survived, fill = survived)) + geom_histogram(alpha = .4, position = position_dodge()) + labs(x = \u0026quot;fare\u0026quot;, y = NULL, color = \u0026quot;survived?\u0026quot;, fill = \u0026quot;survived?\u0026quot;) dataset %\u0026gt;% ggplot(aes(familyage, group = survived, color = survived, fill = survived)) + geom_histogram(alpha = .4, position = position_dodge()) + labs(x = \u0026quot;family aged\u0026quot;, y = NULL, color = \u0026quot;survived?\u0026quot;, fill = \u0026quot;survived?\u0026quot;) Training/testing datasets Split our dataset in two, one dataset for training and the other one for testing. We will use an additionnal splitting step for cross-validation.\nset.seed(2021) spl \u0026lt;- initial_split(dataset, strata = \u0026quot;survived\u0026quot;) train \u0026lt;- training(spl) test \u0026lt;- testing(spl) train_5fold \u0026lt;- train %\u0026gt;% vfold_cv(5) Gradient boosting algorithms - xgboost Let’s start with gradient boosting methods which are very popular in the ML community.\nTuning Set up defaults.\nmset \u0026lt;- metric_set(accuracy) # metric is accuracy control \u0026lt;- control_grid(save_workflow = TRUE, save_pred = TRUE, extract = extract_model) # grid for tuning First a recipe.\nxg_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median step_dummy(all_nominal_predictors()) # all factors var are split into binary terms (factor disj coding) Then specify a gradient boosting model.\nxg_model \u0026lt;- boost_tree(mode = \u0026quot;classification\u0026quot;, # binary response trees = tune(), mtry = tune(), tree_depth = tune(), learn_rate = tune(), loss_reduction = tune(), min_n = tune()) # parameters to be tuned Now set our workflow.\nxg_wf \u0026lt;- workflow() %\u0026gt;% add_model(xg_model) %\u0026gt;% add_recipe(xg_rec) Use cross-validation to evaluate our model with different param config.\nxg_tune \u0026lt;- xg_wf %\u0026gt;% tune_grid(train_5fold, metrics = mset, control = control, grid = crossing(trees = 1000, mtry = c(3, 5, 8), # finalize(mtry(), train) tree_depth = c(5, 10, 15), learn_rate = c(0.01, 0.005), loss_reduction = c(0.01, 0.1, 1), min_n = c(2, 10, 25))) Visualize the results.\nautoplot(xg_tune) + theme_light() Collect metrics.\nxg_tune %\u0026gt;% collect_metrics() %\u0026gt;% arrange(desc(mean)) ## # A tibble: 162 × 12 ## mtry trees min_n tree_depth learn_rate loss_reduction .metric .estimator mean n ## \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;int\u0026gt; ## 1 3 1000 2 5 0.01 0.01 accuracy binary 0.849 5 ## 2 8 1000 2 5 0.01 0.01 accuracy binary 0.847 5 ## 3 8 1000 2 5 0.01 0.1 accuracy binary 0.846 5 ## 4 3 1000 2 15 0.01 0.1 accuracy binary 0.844 5 ## 5 5 1000 2 10 0.01 1 accuracy binary 0.844 5 ## 6 3 1000 2 5 0.01 0.1 accuracy binary 0.844 5 ## 7 5 1000 2 10 0.01 0.1 accuracy binary 0.843 5 ## 8 3 1000 2 10 0.01 0.1 accuracy binary 0.843 5 ## 9 5 1000 2 5 0.01 0.01 accuracy binary 0.843 5 ## 10 5 1000 2 5 0.01 0.1 accuracy binary 0.843 5 ## # … with 152 more rows, and 2 more variables: std_err \u0026lt;dbl\u0026gt;, .config \u0026lt;chr\u0026gt; The tuning takes some time. There are other ways to explore the parameter space more efficiently. For example, we will use the function dials::grid_max_entropy() in the last section about ensemble modelling. Here, I will use finetune::tune_race_anova.\nlibrary(finetune) xg_tune \u0026lt;- xg_wf %\u0026gt;% tune_race_anova( train_5fold, grid = 50, param_info = xg_model %\u0026gt;% parameters(), metrics = metric_set(accuracy), control = control_race(verbose_elim = TRUE)) Visualize the results.\nautoplot(xg_tune) Collect metrics.\nxg_tune %\u0026gt;% collect_metrics() %\u0026gt;% arrange(desc(mean)) ## # A tibble: 50 × 12 ## mtry trees min_n tree_depth learn_rate loss_reduction .metric .estimator mean n ## \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;int\u0026gt; ## 1 6 856 4 13 1.12e- 2 2.49e- 8 accuracy binary 0.837 5 ## 2 10 1952 6 7 3.36e- 2 2.07e+ 0 accuracy binary 0.829 5 ## 3 3 896 2 5 1.73e- 5 6.97e- 8 accuracy binary 0.826 5 ## 4 14 1122 4 6 1.16e- 6 3.44e+ 0 accuracy binary 0.815 4 ## 5 7 939 8 4 2.88e- 5 1.50e- 4 accuracy binary 0.813 4 ## 6 7 17 10 7 4.54e- 6 6.38e- 3 accuracy binary 0.813 4 ## 7 8 92 9 11 3.60e- 3 3.01e-10 accuracy binary 0.811 4 ## 8 13 1407 15 4 1.48e- 2 9.68e- 4 accuracy binary 0.807 3 ## 9 4 658 11 9 9.97e-10 1.27e- 5 accuracy binary 0.805 3 ## 10 2 628 14 9 1.84e- 6 9.44e- 5 accuracy binary 0.798 3 ## # … with 40 more rows, and 2 more variables: std_err \u0026lt;dbl\u0026gt;, .config \u0026lt;chr\u0026gt; Fit model Use best config to fit model to training data.\nxg_fit \u0026lt;- xg_wf %\u0026gt;% finalize_workflow(select_best(xg_tune)) %\u0026gt;% fit(train) ## [23:39:25] WARNING: amalgamation/../src/learner.cc:1095: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior. Check out accuracy on testing dataset to see if we overfitted.\nxg_fit %\u0026gt;% augment(test, type.predict = \u0026quot;response\u0026quot;) %\u0026gt;% accuracy(survived, .pred_class) ## # A tibble: 1 × 3 ## .metric .estimator .estimate ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; ## 1 accuracy binary 0.812 Check out important features (aka predictors).\nimportances \u0026lt;- xgboost::xgb.importance(model = extract_fit_engine(xg_fit)) importances %\u0026gt;% mutate(Feature = fct_reorder(Feature, Gain)) %\u0026gt;% ggplot(aes(Gain, Feature)) + geom_col() Make predictions Now we’re ready to predict survival for the holdout dataset and submit to Kaggle. Note that I use the whole dataset, not just the training dataset.\nxg_wf %\u0026gt;% finalize_workflow(select_best(xg_tune)) %\u0026gt;% fit(dataset) %\u0026gt;% augment(holdout) %\u0026gt;% select(PassengerId, Survived = .pred_class) %\u0026gt;% mutate(Survived = if_else(Survived == \u0026quot;yes\u0026quot;, 1, 0)) %\u0026gt;% write_csv(\u0026quot;output/titanic/xgboost.csv\u0026quot;) ## [23:39:28] WARNING: amalgamation/../src/learner.cc:1095: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior. I got and accuracy of 0.74162. Cool. Let’s train a random forest model now.\nRandom forests Let’s continue with random forest methods.\nTuning First a recipe.\nrf_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median step_dummy(all_nominal_predictors()) # all factors var are split into binary terms (factor disj coding) Then specify a random forest model.\nrf_model \u0026lt;- rand_forest(mode = \u0026quot;classification\u0026quot;, # binary response engine = \u0026quot;ranger\u0026quot;, # by default mtry = tune(), trees = tune(), min_n = tune()) # parameters to be tuned Now set our workflow.\nrf_wf \u0026lt;- workflow() %\u0026gt;% add_model(rf_model) %\u0026gt;% add_recipe(rf_rec) Use cross-validation to evaluate our model with different param config.\nrf_tune \u0026lt;- rf_wf %\u0026gt;% tune_race_anova( train_5fold, grid = 50, param_info = rf_model %\u0026gt;% parameters(), metrics = metric_set(accuracy), control = control_race(verbose_elim = TRUE)) Visualize the results.\nautoplot(rf_tune) Collect metrics.\nrf_tune %\u0026gt;% collect_metrics() %\u0026gt;% arrange(desc(mean)) ## # A tibble: 50 × 9 ## mtry trees min_n .metric .estimator mean n std_err .config ## \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; ## 1 14 1554 5 accuracy binary 0.837 5 0.00656 Preprocessor1_Model49 ## 2 4 774 18 accuracy binary 0.837 5 0.0133 Preprocessor1_Model50 ## 3 5 1736 8 accuracy binary 0.834 5 0.0111 Preprocessor1_Model46 ## 4 8 1322 5 accuracy binary 0.832 5 0.00713 Preprocessor1_Model41 ## 5 2 1078 30 accuracy binary 0.831 5 0.00727 Preprocessor1_Model12 ## 6 4 1892 14 accuracy binary 0.831 5 0.00886 Preprocessor1_Model39 ## 7 8 962 7 accuracy binary 0.829 5 0.00742 Preprocessor1_Model43 ## 8 7 946 4 accuracy binary 0.826 5 0.00452 Preprocessor1_Model23 ## 9 7 1262 3 accuracy binary 0.826 5 0.00710 Preprocessor1_Model47 ## 10 9 544 9 accuracy binary 0.825 5 0.00673 Preprocessor1_Model11 ## # … with 40 more rows Fit model Use best config to fit model to training data.\nrf_fit \u0026lt;- rf_wf %\u0026gt;% finalize_workflow(select_best(rf_tune)) %\u0026gt;% fit(train) Check out accuracy on testing dataset to see if we overfitted.\nrf_fit %\u0026gt;% augment(test, type.predict = \u0026quot;response\u0026quot;) %\u0026gt;% accuracy(survived, .pred_class) ## # A tibble: 1 × 3 ## .metric .estimator .estimate ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; ## 1 accuracy binary 0.786 Check out important features (aka predictors).\nlibrary(vip) finalize_model( x = rf_model, parameters = select_best(rf_tune)) %\u0026gt;% set_engine(\u0026quot;ranger\u0026quot;, importance = \u0026quot;permutation\u0026quot;) %\u0026gt;% fit(survived ~ ., data = juice(prep(rf_rec))) %\u0026gt;% vip(geom = \u0026quot;point\u0026quot;) Make predictions Now we’re ready to predict survival for the holdout dataset and submit to Kaggle.\nrf_wf %\u0026gt;% finalize_workflow(select_best(rf_tune)) %\u0026gt;% fit(dataset) %\u0026gt;% augment(holdout) %\u0026gt;% select(PassengerId, Survived = .pred_class) %\u0026gt;% mutate(Survived = if_else(Survived == \u0026quot;yes\u0026quot;, 1, 0)) %\u0026gt;% write_csv(\u0026quot;output/titanic/randomforest.csv\u0026quot;) I got and accuracy of 0.77990, a bit better than gradient boosting.\nLet’s continue with cat boosting methods.\nGradient boosting algorithms - catboost Tuning Set up defaults.\nmset \u0026lt;- metric_set(accuracy) # metric is accuracy control \u0026lt;- control_grid(save_workflow = TRUE, save_pred = TRUE, extract = extract_model) # grid for tuning First a recipe.\ncb_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median step_dummy(all_nominal_predictors()) # all factors var are split into binary terms (factor disj coding) Then specify a cat boosting model.\nlibrary(treesnip) cb_model \u0026lt;- boost_tree(mode = \u0026quot;classification\u0026quot;, engine = \u0026quot;catboost\u0026quot;, mtry = tune(), trees = tune(), min_n = tune(), tree_depth = tune(), learn_rate = tune()) # parameters to be tuned Now set our workflow.\ncb_wf \u0026lt;- workflow() %\u0026gt;% add_model(cb_model) %\u0026gt;% add_recipe(cb_rec) Use cross-validation to evaluate our model with different param config.\ncb_tune \u0026lt;- cb_wf %\u0026gt;% tune_race_anova( train_5fold, grid = 30, param_info = cb_model %\u0026gt;% parameters(), metrics = metric_set(accuracy), control = control_race(verbose_elim = TRUE)) Visualize the results.\nautoplot(cb_tune) Collect metrics.\ncb_tune %\u0026gt;% collect_metrics() %\u0026gt;% arrange(desc(mean)) ## # A tibble: 30 × 11 ## mtry trees min_n tree_depth learn_rate .metric .estimator mean n std_err .config ## \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; ## 1 1 1787 25 7 6.84e- 3 accuracy binary 0.835 5 0.0125 Preproc… ## 2 12 1885 18 15 2.06e- 3 accuracy binary 0.831 5 0.00602 Preproc… ## 3 13 1278 5 6 2.10e- 2 accuracy binary 0.826 5 0.00431 Preproc… ## 4 3 1681 22 5 5.42e- 3 accuracy binary 0.825 5 0.00507 Preproc… ## 5 9 303 2 8 9.94e- 2 accuracy binary 0.820 4 0.0120 Preproc… ## 6 11 1201 24 12 3.77e- 2 accuracy binary 0.812 3 0.00868 Preproc… ## 7 11 634 35 11 1.30e- 3 accuracy binary 0.805 3 0.0242 Preproc… ## 8 7 1648 4 7 2.38e- 4 accuracy binary 0.737 3 0.0115 Preproc… ## 9 1 1427 27 10 1.74e- 6 accuracy binary 0.378 3 0.0152 Preproc… ## 10 2 940 19 3 3.20e-10 accuracy binary 0.378 3 0.0152 Preproc… ## # … with 20 more rows Fit model Use best config to fit model to training data.\ncb_fit \u0026lt;- cb_wf %\u0026gt;% finalize_workflow(select_best(cb_tune)) %\u0026gt;% fit(train) Check out accuracy on testing dataset to see if we overfitted.\ncb_fit %\u0026gt;% augment(test, type.predict = \u0026quot;response\u0026quot;) %\u0026gt;% accuracy(survived, .pred_class) ## # A tibble: 1 × 3 ## .metric .estimator .estimate ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; ## 1 accuracy binary 0.808 Make predictions Now we’re ready to predict survival for the holdout dataset and submit to Kaggle.\ncb_wf %\u0026gt;% finalize_workflow(select_best(cb_tune)) %\u0026gt;% fit(dataset) %\u0026gt;% augment(holdout) %\u0026gt;% select(PassengerId, Survived = .pred_class) %\u0026gt;% mutate(Survived = if_else(Survived == \u0026quot;yes\u0026quot;, 1, 0)) %\u0026gt;% write_csv(\u0026quot;output/titanic/catboost.csv\u0026quot;) I got and accuracy of 0.76076. Cool.\nRegularization methods Let’s continue with elastic net regularization.\nTuning First a recipe.\nen_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median step_normalize(all_numeric_predictors()) %\u0026gt;% # normalize step_dummy(all_nominal_predictors()) # all factors var are split into binary terms (factor disj coding) Then specify a regularization model. We tune parameter mixture, with ridge regression for mixture = 0, and lasso for mixture = 1.\nen_model \u0026lt;- logistic_reg(penalty = tune(), mixture = tune()) %\u0026gt;% # param to be tuned set_engine(\u0026quot;glmnet\u0026quot;) %\u0026gt;% # elastic net set_mode(\u0026quot;classification\u0026quot;) # binary response Now set our workflow.\nen_wf \u0026lt;- workflow() %\u0026gt;% add_model(en_model) %\u0026gt;% add_recipe(en_rec) Use cross-validation to evaluate our model with different param config.\nen_tune \u0026lt;- en_wf %\u0026gt;% tune_grid(train_5fold, metrics = mset, control = control, grid = crossing(penalty = 10 ^ seq(-8, -.5, .5), mixture = seq(0, 1, length.out = 10))) Visualize the results.\nautoplot(en_tune) Collect metrics.\nen_tune %\u0026gt;% collect_metrics() %\u0026gt;% arrange(desc(mean)) ## # A tibble: 160 × 8 ## penalty mixture .metric .estimator mean n std_err .config ## \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; ## 1 0.00000001 0.111 accuracy binary 0.831 5 0.0112 Preprocessor1_Model017 ## 2 0.0000000316 0.111 accuracy binary 0.831 5 0.0112 Preprocessor1_Model018 ## 3 0.0000001 0.111 accuracy binary 0.831 5 0.0112 Preprocessor1_Model019 ## 4 0.000000316 0.111 accuracy binary 0.831 5 0.0112 Preprocessor1_Model020 ## 5 0.000001 0.111 accuracy binary 0.831 5 0.0112 Preprocessor1_Model021 ## 6 0.00000316 0.111 accuracy binary 0.831 5 0.0112 Preprocessor1_Model022 ## 7 0.00001 0.111 accuracy binary 0.831 5 0.0112 Preprocessor1_Model023 ## 8 0.0000316 0.111 accuracy binary 0.831 5 0.0112 Preprocessor1_Model024 ## 9 0.0001 0.111 accuracy binary 0.831 5 0.0112 Preprocessor1_Model025 ## 10 0.000316 0.111 accuracy binary 0.831 5 0.0112 Preprocessor1_Model026 ## # … with 150 more rows Fit model Use best config to fit model to training data.\nen_fit \u0026lt;- en_wf %\u0026gt;% finalize_workflow(select_best(en_tune)) %\u0026gt;% fit(train) Check out accuracy on testing dataset to see if we overfitted.\nen_fit %\u0026gt;% augment(test, type.predict = \u0026quot;response\u0026quot;) %\u0026gt;% accuracy(survived, .pred_class) ## # A tibble: 1 × 3 ## .metric .estimator .estimate ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; ## 1 accuracy binary 0.826 Check out important features (aka predictors).\nlibrary(broom) en_fit$fit$fit$fit %\u0026gt;% tidy() %\u0026gt;% filter(lambda \u0026gt;= select_best(en_tune)$penalty) %\u0026gt;% filter(lambda == min(lambda), term != \u0026quot;(Intercept)\u0026quot;) %\u0026gt;% mutate(term = fct_reorder(term, estimate)) %\u0026gt;% ggplot(aes(estimate, term, fill = estimate \u0026gt; 0)) + geom_col() + theme(legend.position = \u0026quot;none\u0026quot;) Make predictions Now we’re ready to predict survival for the holdout dataset and submit to Kaggle.\nen_wf %\u0026gt;% finalize_workflow(select_best(en_tune)) %\u0026gt;% fit(dataset) %\u0026gt;% augment(holdout) %\u0026gt;% select(PassengerId, Survived = .pred_class) %\u0026gt;% mutate(Survived = if_else(Survived == \u0026quot;yes\u0026quot;, 1, 0)) %\u0026gt;% write_csv(\u0026quot;output/titanic/regularization.csv\u0026quot;) I got and accuracy of 0.76315.\nLogistic regression And what about a good old-fashioned logistic regression (not a ML algo)?\nFirst a recipe.\nlogistic_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median step_normalize(all_numeric_predictors()) %\u0026gt;% # normalize step_dummy(all_nominal_predictors()) # all factors var are split into binary terms (factor disj coding) Then specify a logistic regression.\nlogistic_model \u0026lt;- logistic_reg() %\u0026gt;% # no param to be tuned set_engine(\u0026quot;glm\u0026quot;) %\u0026gt;% # elastic net set_mode(\u0026quot;classification\u0026quot;) # binary response Now set our workflow.\nlogistic_wf \u0026lt;- workflow() %\u0026gt;% add_model(logistic_model) %\u0026gt;% add_recipe(logistic_rec) Fit model.\nlogistic_fit \u0026lt;- logistic_wf %\u0026gt;% fit(train) Inspect significant features (aka predictors).\ntidy(logistic_fit, exponentiate = TRUE) %\u0026gt;% filter(p.value \u0026lt; 0.05) ## # A tibble: 6 × 5 ## term estimate std.error statistic p.value ## \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; ## 1 age 1.35 0.152 1.98 0.0473 ## 2 familyage 2.97 0.223 4.88 0.00000105 ## 3 class_first 0.136 0.367 -5.43 0.0000000559 ## 4 class_second 0.386 0.298 -3.19 0.00145 ## 5 title_Mr. 51.0 0.684 5.75 0.00000000912 ## 6 title_Other 54.7 0.991 4.04 0.0000538 Same thing, but graphically.\nlibrary(broom) logistic_fit %\u0026gt;% tidy() %\u0026gt;% mutate(term = fct_reorder(term, estimate)) %\u0026gt;% ggplot(aes(estimate, term, fill = estimate \u0026gt; 0)) + geom_col() + theme(legend.position = \u0026quot;none\u0026quot;) Check out accuracy on testing dataset to see if we overfitted.\nlogistic_fit %\u0026gt;% augment(test, type.predict = \u0026quot;response\u0026quot;) %\u0026gt;% accuracy(survived, .pred_class) ## # A tibble: 1 × 3 ## .metric .estimator .estimate ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; ## 1 accuracy binary 0.821 Confusion matrix.\nlogistic_fit %\u0026gt;% augment(test, type.predict = \u0026quot;response\u0026quot;) %\u0026gt;% conf_mat(survived, .pred_class) ## Truth ## Prediction yes no ## yes 59 13 ## no 27 125 ROC curve.\nlogistic_fit %\u0026gt;% augment(test, type.predict = \u0026quot;response\u0026quot;) %\u0026gt;% roc_curve(truth = survived, estimate = .pred_yes) %\u0026gt;% autoplot() Now we’re ready to predict survival for the holdout dataset and submit to Kaggle.\nlogistic_wf %\u0026gt;% fit(dataset) %\u0026gt;% augment(holdout) %\u0026gt;% select(PassengerId, Survived = .pred_class) %\u0026gt;% mutate(Survived = if_else(Survived == \u0026quot;yes\u0026quot;, 1, 0)) %\u0026gt;% write_csv(\u0026quot;output/titanic/logistic.csv\u0026quot;) I got and accuracy of 0.76076. Oldies but goodies!\nNeural networks We go on with neural networks.\nTuning Set up defaults.\nmset \u0026lt;- metric_set(accuracy) # metric is accuracy control \u0026lt;- control_grid(save_workflow = TRUE, save_pred = TRUE, extract = extract_model) # grid for tuning First a recipe.\nnn_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median step_normalize(all_numeric_predictors()) %\u0026gt;% step_dummy(all_nominal_predictors()) # all factors var are split into binary terms (factor disj coding) Then specify a neural network.\nnn_model \u0026lt;- mlp(epochs = tune(), hidden_units = tune(), dropout = tune()) %\u0026gt;% # param to be tuned set_mode(\u0026quot;classification\u0026quot;) %\u0026gt;% # binary response var set_engine(\u0026quot;keras\u0026quot;, verbose = 0) Now set our workflow.\nnn_wf \u0026lt;- workflow() %\u0026gt;% add_model(nn_model) %\u0026gt;% add_recipe(nn_rec) Use cross-validation to evaluate our model with different param config.\nnn_tune \u0026lt;- nn_wf %\u0026gt;% tune_race_anova( train_5fold, grid = 30, param_info = nn_model %\u0026gt;% parameters(), metrics = metric_set(accuracy), control = control_race(verbose_elim = TRUE)) Visualize the results.\nautoplot(nn_tune) Collect metrics.\nnn_tune %\u0026gt;% collect_metrics() %\u0026gt;% arrange(desc(mean)) ## # A tibble: 30 × 9 ## hidden_units dropout epochs .metric .estimator mean n std_err .config ## \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;int\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; ## 1 10 0.484 405 accuracy binary 0.838 5 0.0150 Preprocessor1_Model… ## 2 7 0.0597 220 accuracy binary 0.834 5 0.00881 Preprocessor1_Model… ## 3 4 0.536 629 accuracy binary 0.834 5 0.0154 Preprocessor1_Model… ## 4 9 0.198 768 accuracy binary 0.832 5 0.0146 Preprocessor1_Model… ## 5 6 0.752 822 accuracy binary 0.832 5 0.0112 Preprocessor1_Model… ## 6 9 0.406 293 accuracy binary 0.831 5 0.0145 Preprocessor1_Model… ## 7 4 0.00445 871 accuracy binary 0.831 5 0.0150 Preprocessor1_Model… ## 8 4 0.293 353 accuracy binary 0.831 5 0.0117 Preprocessor1_Model… ## 9 10 0.675 935 accuracy binary 0.831 5 0.0141 Preprocessor1_Model… ## 10 7 0.128 580 accuracy binary 0.831 5 0.0151 Preprocessor1_Model… ## # … with 20 more rows Fit model Use best config to fit model to training data.\nnn_fit \u0026lt;- nn_wf %\u0026gt;% finalize_workflow(select_best(nn_tune)) %\u0026gt;% fit(train) Check out accuracy on testing dataset to see if we overfitted.\nnn_fit %\u0026gt;% augment(test, type.predict = \u0026quot;response\u0026quot;) %\u0026gt;% accuracy(survived, .pred_class) ## # A tibble: 1 × 3 ## .metric .estimator .estimate ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; ## 1 accuracy binary 0.817 Make predictions Now we’re ready to predict survival for the holdout dataset and submit to Kaggle.\nnn_wf %\u0026gt;% finalize_workflow(select_best(nn_tune)) %\u0026gt;% fit(train) %\u0026gt;% augment(holdout) %\u0026gt;% select(PassengerId, Survived = .pred_class) %\u0026gt;% mutate(Survived = if_else(Survived == \u0026quot;yes\u0026quot;, 1, 0)) %\u0026gt;% write_csv(\u0026quot;output/titanic/nn.csv\u0026quot;) I got and accuracy of 0.78708. My best score so far.\nSupport vector machines We go on with support vector machines.\nTuning Set up defaults.\nmset \u0026lt;- metric_set(accuracy) # metric is accuracy control \u0026lt;- control_grid(save_workflow = TRUE, save_pred = TRUE, extract = extract_model) # grid for tuning First a recipe.\nsvm_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median # remove any zero variance predictors step_zv(all_predictors()) %\u0026gt;% # remove any linear combinations step_lincomb(all_numeric()) %\u0026gt;% step_dummy(all_nominal_predictors()) # all factors var are split into binary terms (factor disj coding) Then specify a svm.\nsvm_model \u0026lt;- svm_rbf(cost = tune(), rbf_sigma = tune()) %\u0026gt;% # param to be tuned set_mode(\u0026quot;classification\u0026quot;) %\u0026gt;% # binary response var set_engine(\u0026quot;kernlab\u0026quot;) Now set our workflow.\nsvm_wf \u0026lt;- workflow() %\u0026gt;% add_model(svm_model) %\u0026gt;% add_recipe(svm_rec) Use cross-validation to evaluate our model with different param config.\nsvm_tune \u0026lt;- svm_wf %\u0026gt;% tune_race_anova( train_5fold, grid = 30, param_info = svm_model %\u0026gt;% parameters(), metrics = metric_set(accuracy), control = control_race(verbose_elim = TRUE)) Visualize the results.\nautoplot(svm_tune) Collect metrics.\nsvm_tune %\u0026gt;% collect_metrics() %\u0026gt;% arrange(desc(mean)) ## # A tibble: 30 × 8 ## cost rbf_sigma .metric .estimator mean n std_err .config ## \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; ## 1 17.7 0.00183 accuracy binary 0.829 5 0.0101 Preprocessor1_Model05 ## 2 0.741 0.0506 accuracy binary 0.823 5 0.0129 Preprocessor1_Model14 ## 3 1.77 0.0429 accuracy binary 0.820 5 0.0143 Preprocessor1_Model08 ## 4 0.229 0.0197 accuracy binary 0.808 3 0.00626 Preprocessor1_Model23 ## 5 1.15 0.00285 accuracy binary 0.795 3 0.00867 Preprocessor1_Model28 ## 6 0.00182 0.00000203 accuracy binary 0.613 3 0.0171 Preprocessor1_Model01 ## 7 0.0477 0.714 accuracy binary 0.613 3 0.0171 Preprocessor1_Model02 ## 8 6.60 0.0000294 accuracy binary 0.613 3 0.0171 Preprocessor1_Model03 ## 9 0.00254 0.000636 accuracy binary 0.613 3 0.0171 Preprocessor1_Model04 ## 10 0.00544 0.0000000647 accuracy binary 0.613 3 0.0171 Preprocessor1_Model06 ## # … with 20 more rows Fit model Use best config to fit model to training data.\nsvm_fit \u0026lt;- svm_wf %\u0026gt;% finalize_workflow(select_best(svm_tune)) %\u0026gt;% fit(train) Check out accuracy on testing dataset to see if we overfitted.\nsvm_fit %\u0026gt;% augment(test, type.predict = \u0026quot;response\u0026quot;) %\u0026gt;% accuracy(survived, .pred_class) ## # A tibble: 1 × 3 ## .metric .estimator .estimate ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; ## 1 accuracy binary 0.826 Make predictions Now we’re ready to predict survival for the holdout dataset and submit to Kaggle.\nsvm_wf %\u0026gt;% finalize_workflow(select_best(svm_tune)) %\u0026gt;% fit(dataset) %\u0026gt;% augment(holdout) %\u0026gt;% select(PassengerId, Survived = .pred_class) %\u0026gt;% mutate(Survived = if_else(Survived == \u0026quot;yes\u0026quot;, 1, 0)) %\u0026gt;% write_csv(\u0026quot;output/titanic/svm.csv\u0026quot;) I got and accuracy of 0.77511.\nDecision trees We go on with decision trees.\nTuning Set up defaults.\nmset \u0026lt;- metric_set(accuracy) # metric is accuracy control \u0026lt;- control_grid(save_workflow = TRUE, save_pred = TRUE, extract = extract_model) # grid for tuning First a recipe.\ndt_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median step_zv(all_predictors()) %\u0026gt;% step_dummy(all_nominal_predictors()) # all factors var are split into binary terms (factor disj coding) Then specify a decision tree model.\nlibrary(baguette) dt_model \u0026lt;- bag_tree(cost_complexity = tune(), tree_depth = tune(), min_n = tune()) %\u0026gt;% # param to be tuned set_engine(\u0026quot;rpart\u0026quot;, times = 25) %\u0026gt;% # nb bootstraps set_mode(\u0026quot;classification\u0026quot;) # binary response var Now set our workflow.\ndt_wf \u0026lt;- workflow() %\u0026gt;% add_model(dt_model) %\u0026gt;% add_recipe(dt_rec) Use cross-validation to evaluate our model with different param config.\ndt_tune \u0026lt;- dt_wf %\u0026gt;% tune_race_anova( train_5fold, grid = 30, param_info = dt_model %\u0026gt;% parameters(), metrics = metric_set(accuracy), control = control_race(verbose_elim = TRUE)) Visualize the results.\nautoplot(dt_tune) Collect metrics.\ndt_tune %\u0026gt;% collect_metrics() %\u0026gt;% arrange(desc(mean)) ## # A tibble: 30 × 9 ## cost_complexity tree_depth min_n .metric .estimator mean n std_err .config ## \u0026lt;dbl\u0026gt; \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; ## 1 3.02e- 7 13 7 accuracy binary 0.832 5 0.0130 Preprocessor1_… ## 2 3.43e- 3 14 4 accuracy binary 0.829 5 0.0145 Preprocessor1_… ## 3 5.75e- 5 5 7 accuracy binary 0.828 5 0.0126 Preprocessor1_… ## 4 2.13e- 6 14 5 accuracy binary 0.828 5 0.0108 Preprocessor1_… ## 5 1.34e- 5 5 35 accuracy binary 0.823 5 0.00654 Preprocessor1_… ## 6 4.47e- 5 13 10 accuracy binary 0.823 5 0.0120 Preprocessor1_… ## 7 1.42e- 2 4 25 accuracy binary 0.822 5 0.0121 Preprocessor1_… ## 8 4.54e-10 10 36 accuracy binary 0.822 5 0.0147 Preprocessor1_… ## 9 8.10e- 8 11 32 accuracy binary 0.822 5 0.0143 Preprocessor1_… ## 10 3.43e- 4 11 21 accuracy binary 0.820 5 0.0207 Preprocessor1_… ## # … with 20 more rows Fit model Use best config to fit model to training data.\ndt_fit \u0026lt;- dt_wf %\u0026gt;% finalize_workflow(select_best(dt_tune)) %\u0026gt;% fit(train) Check out accuracy on testing dataset to see if we overfitted.\ndt_fit %\u0026gt;% augment(test, type.predict = \u0026quot;response\u0026quot;) %\u0026gt;% accuracy(survived, .pred_class) ## # A tibble: 1 × 3 ## .metric .estimator .estimate ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; ## 1 accuracy binary 0.808 Make predictions Now we’re ready to predict survival for the holdout dataset and submit to Kaggle.\ndt_wf %\u0026gt;% finalize_workflow(select_best(dt_tune)) %\u0026gt;% fit(dataset) %\u0026gt;% augment(holdout) %\u0026gt;% select(PassengerId, Survived = .pred_class) %\u0026gt;% mutate(Survived = if_else(Survived == \u0026quot;yes\u0026quot;, 1, 0)) %\u0026gt;% write_csv(\u0026quot;output/titanic/dt.csv\u0026quot;) I got and accuracy of 0.76794.\nStacked ensemble modelling Let’s do some ensemble modelling with all algo but logistic and catboost. Tune again with a probability-based metric. Start with xgboost.\nlibrary(finetune) library(stacks) # xgboost xg_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median step_dummy(all_nominal_predictors()) # all factors var are split into binary terms (factor disj coding) xg_model \u0026lt;- boost_tree(mode = \u0026quot;classification\u0026quot;, # binary response trees = tune(), mtry = tune(), tree_depth = tune(), learn_rate = tune(), loss_reduction = tune(), min_n = tune()) # parameters to be tuned xg_wf \u0026lt;- workflow() %\u0026gt;% add_model(xg_model) %\u0026gt;% add_recipe(xg_rec) xg_grid \u0026lt;- grid_latin_hypercube( trees(), finalize(mtry(), train), tree_depth(), learn_rate(), loss_reduction(), min_n(), size = 30) xg_tune \u0026lt;- xg_wf %\u0026gt;% tune_grid( resamples = train_5fold, grid = xg_grid, metrics = metric_set(roc_auc), control = control_stack_grid()) Then random forests.\n# random forest rf_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% step_dummy(all_nominal_predictors()) rf_model \u0026lt;- rand_forest(mode = \u0026quot;classification\u0026quot;, # binary response engine = \u0026quot;ranger\u0026quot;, # by default mtry = tune(), trees = tune(), min_n = tune()) # parameters to be tuned rf_wf \u0026lt;- workflow() %\u0026gt;% add_model(rf_model) %\u0026gt;% add_recipe(rf_rec) rf_grid \u0026lt;- grid_latin_hypercube( finalize(mtry(), train), trees(), min_n(), size = 30) rf_tune \u0026lt;- rf_wf %\u0026gt;% tune_grid( resamples = train_5fold, grid = rf_grid, metrics = metric_set(roc_auc), control = control_stack_grid()) Regularisation methods (between ridge and lasso).\n# regularization methods en_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median step_normalize(all_numeric_predictors()) %\u0026gt;% # normalize step_dummy(all_nominal_predictors()) en_model \u0026lt;- logistic_reg(penalty = tune(), mixture = tune()) %\u0026gt;% # param to be tuned set_engine(\u0026quot;glmnet\u0026quot;) %\u0026gt;% # elastic net set_mode(\u0026quot;classification\u0026quot;) # binary response en_wf \u0026lt;- workflow() %\u0026gt;% add_model(en_model) %\u0026gt;% add_recipe(en_rec) en_grid \u0026lt;- grid_latin_hypercube( penalty(), mixture(), size = 30) en_tune \u0026lt;- en_wf %\u0026gt;% tune_grid( resamples = train_5fold, grid = en_grid, metrics = metric_set(roc_auc), control = control_stack_grid()) Neural networks (takes time, so pick only a few values for illustration purpose).\n# neural networks nn_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median step_normalize(all_numeric_predictors()) %\u0026gt;% step_dummy(all_nominal_predictors()) nn_model \u0026lt;- mlp(epochs = tune(), hidden_units = 2, dropout = tune()) %\u0026gt;% # param to be tuned set_mode(\u0026quot;classification\u0026quot;) %\u0026gt;% # binary response var set_engine(\u0026quot;keras\u0026quot;, verbose = 0) nn_wf \u0026lt;- workflow() %\u0026gt;% add_model(nn_model) %\u0026gt;% add_recipe(nn_rec) # nn_grid \u0026lt;- grid_latin_hypercube( # epochs(), # hidden_units(), # dropout(), # size = 10) nn_tune \u0026lt;- nn_wf %\u0026gt;% tune_grid( resamples = train_5fold, grid = crossing(dropout = c(0.1, 0.2), epochs = c(250, 500, 1000)), # nn_grid metrics = metric_set(roc_auc), control = control_stack_grid()) #autoplot(nn_tune) Support vector machines.\n# support vector machines svm_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% step_normalize(all_numeric_predictors()) %\u0026gt;% step_dummy(all_nominal_predictors()) svm_model \u0026lt;- svm_rbf(cost = tune(), rbf_sigma = tune()) %\u0026gt;% # param to be tuned set_mode(\u0026quot;classification\u0026quot;) %\u0026gt;% # binary response var set_engine(\u0026quot;kernlab\u0026quot;) svm_wf \u0026lt;- workflow() %\u0026gt;% add_model(svm_model) %\u0026gt;% add_recipe(svm_rec) svm_grid \u0026lt;- grid_latin_hypercube( cost(), rbf_sigma(), size = 30) svm_tune \u0026lt;- svm_wf %\u0026gt;% tune_grid( resamples = train_5fold, grid = svm_grid, metrics = metric_set(roc_auc), control = control_stack_grid()) Last, decision trees.\n# decision trees dt_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% step_zv(all_predictors()) %\u0026gt;% step_dummy(all_nominal_predictors()) library(baguette) dt_model \u0026lt;- bag_tree(cost_complexity = tune(), tree_depth = tune(), min_n = tune()) %\u0026gt;% # param to be tuned set_engine(\u0026quot;rpart\u0026quot;, times = 25) %\u0026gt;% # nb bootstraps set_mode(\u0026quot;classification\u0026quot;) # binary response var dt_wf \u0026lt;- workflow() %\u0026gt;% add_model(dt_model) %\u0026gt;% add_recipe(dt_rec) dt_grid \u0026lt;- grid_latin_hypercube( cost_complexity(), tree_depth(), min_n(), size = 30) dt_tune \u0026lt;- dt_wf %\u0026gt;% tune_grid( resamples = train_5fold, grid = dt_grid, metrics = metric_set(roc_auc), control = control_stack_grid()) Get best config.\nxg_best \u0026lt;- xg_tune %\u0026gt;% filter_parameters(parameters = select_best(xg_tune)) rf_best \u0026lt;- rf_tune %\u0026gt;% filter_parameters(parameters = select_best(rf_tune)) en_best \u0026lt;- en_tune %\u0026gt;% filter_parameters(parameters = select_best(en_tune)) nn_best \u0026lt;- nn_tune %\u0026gt;% filter_parameters(parameters = select_best(nn_tune)) svm_best \u0026lt;- svm_tune %\u0026gt;% filter_parameters(parameters = select_best(svm_tune)) dt_best \u0026lt;- dt_tune %\u0026gt;% filter_parameters(parameters = select_best(dt_tune)) Do the stacked ensemble modelling.\nPile all models together.\nblended \u0026lt;- stacks() %\u0026gt;% # initialize add_candidates(en_best) %\u0026gt;% # add regularization model add_candidates(xg_best) %\u0026gt;% # add gradient boosting add_candidates(rf_best) %\u0026gt;% # add random forest add_candidates(nn_best) %\u0026gt;% # add neural network add_candidates(svm_best) %\u0026gt;% # add svm add_candidates(dt_best) # add decision trees blended ## # A data stack with 6 model definitions and 6 candidate members: ## # en_best: 1 model configuration ## # xg_best: 1 model configuration ## # rf_best: 1 model configuration ## # nn_best: 1 model configuration ## # svm_best: 1 model configuration ## # dt_best: 1 model configuration ## # Outcome: survived (factor) Fit regularized model.\nblended_fit \u0026lt;- blended %\u0026gt;% blend_predictions() # fit regularized model Visualise penalized model. Note that neural networks are dropped, despite achieving best score when used in isolation. I’ll have to dig into that.\nautoplot(blended_fit) autoplot(blended_fit, type = \u0026quot;members\u0026quot;) autoplot(blended_fit, type = \u0026quot;weights\u0026quot;) Fit candidate members with non-zero stacking coef with full training dataset.\nblended_regularized \u0026lt;- blended_fit %\u0026gt;% fit_members() blended_regularized ## # A tibble: 3 × 3 ## member type weight ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; ## 1 .pred_no_en_best_1_12 logistic_reg 2.67 ## 2 .pred_no_dt_best_1_11 bag_tree 1.95 ## 3 .pred_no_rf_best_1_05 rand_forest 0.922 Perf on testing dataset?\ntest %\u0026gt;% bind_cols(predict(blended_regularized, .)) %\u0026gt;% accuracy(survived, .pred_class) ## # A tibble: 1 × 3 ## .metric .estimator .estimate ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; ## 1 accuracy binary 0.826 Now predict.\nholdout %\u0026gt;% bind_cols(predict(blended_regularized, .)) %\u0026gt;% select(PassengerId, Survived = .pred_class) %\u0026gt;% mutate(Survived = if_else(Survived == \u0026quot;yes\u0026quot;, 1, 0)) %\u0026gt;% write_csv(\u0026quot;output/titanic/stacked.csv\u0026quot;) I got an 0.76076 accuracy.\nConclusions I covered several ML algorithms and logistic regression with the awesome tidymodels metapackage in R. My scores at predicting Titanic survivors were ok I guess. Some folks on Kaggle got a perfect accuracy, so there is always room for improvement. Maybe better tuning, better features (or predictors) or other algorithms would increase accuracy. Of course, I forgot to use set.seed() so results are not exactly reproducible.\nThis post was also published on https://www.r-bloggers.com.\u0026#160;\u0026#x21a9;\u0026#xfe0e;\n","date":1628812800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1628812800,"objectID":"6d1999265d6ae3cfa97cff110ffd9567","permalink":"https://oliviergimenez.github.io/blog/learning-machine-learning/","publishdate":"2021-08-13T00:00:00Z","relpermalink":"/blog/learning-machine-learning/","section":"blog","summary":"I would like to familiarize myself with machine learning (ML) techniques in R. So I have been reading and learning by doing. I thought I\u0026rsquo;d share my experience for others who\u0026rsquo;d like to give it a try1.\n","tags":["machine learning","tidymodels","R","rstats"],"title":"Experimenting with machine learning in R with tidymodels and the Kaggle titanic dataset","type":"blog"},{"authors":null,"categories":null,"content":"I use R to retrieve some data from Twitter, do some exploratory data analysis and visualisation and examine a network of followers.\nMotivation I use Twitter to get live updates of what my follow scientists are up to, to communicate about my students\u0026rsquo; awesome work and to share material that I hope is useful to some people1.\nRecently, I reached 5,000 followers and I thought I\u0026rsquo;d spend some time trying to know better who they/you are. To do so, I use R to retrieve some data from Twitter using rtweet, do some data exploration and visualisation using the tidyverse and examine my network of followers with tidygraph, ggraph and igraph. Data and codes are available from https://github.com/oliviergimenez/sna-twitter.\nTo reproduce the analyses below, you will need to access a Twitter API (application programming interface) to retrieve the information about your followers. In brief, an API is an intermediary application that allows applications to talk to each other. To access the Twitter APIs, you need a developer account for which you may apply at https://developer.twitter.com/en. There is a short form to fill in, and it takes less than a day to get an answer.\nBelow I rely heavily on the code shared by Joe Cristian through the Algoritma Technical Blog at https://algotech.netlify.app/blog/social-network-analysis-in-r/. Kuddos and credits to him.\nData retrieving We load the rtweet package to work with Twitter from R.\n#devtools::install_github(\u0026quot;ropensci/rtweet\u0026quot;) library(rtweet) We also load the tidyverse for data manipulation and visualisation.\nlibrary(tidyverse) theme_set(theme_light(base_size = 14)) First we need to get credentials. The usual rtweet sequence with rtweet_app, auth_save and auth_as is supposed to work (see here), but the Twitter API kept failing (error 401) for me. Tried a few things, in vain. I will use the deprecated create_token function instead. You might need to change the defaults of your Twitter app from \u0026ldquo;Read only\u0026rdquo; to \u0026ldquo;Read, write and access direct messages\u0026rdquo;.\nEnter my API keys and access tokens (not shown), then authenticate.\ntoken \u0026lt;- create_token( app = \u0026quot;sna-twitter-network-5k\u0026quot;, consumer_key = api_key, consumer_secret = api_secret_key, access_token = access_token, access_secret = access_token_secret) Get my Twitter info with my description, number of followers, number of likes, etc.\nog \u0026lt;- lookup_users(\u0026quot;oaggimenez\u0026quot;) str(og, max = 2) ## 'data.frame':\t1 obs. of 21 variables: ## $ id : num 7.51e+17 ## $ id_str : chr \u0026quot;750892662224453632\u0026quot; ## $ name : chr \u0026quot;Olivier Gimenez \\U0001f596\u0026quot; ## $ screen_name : chr \u0026quot;oaggimenez\u0026quot; ## $ location : chr \u0026quot;Montpellier, France\u0026quot; ## $ description : chr \u0026quot;Scientist (he/him) @CNRS @cefemontpellier @twitthair1 • Grown statistician • Improvised ecologist • Sociologist\u0026quot;| __truncated__ ## $ url : chr \u0026quot;https://t.co/l7NImYeGdY\u0026quot; ## $ protected : logi FALSE ## $ followers_count : int 5042 ## $ friends_count : int 2191 ## $ listed_count : int 68 ## $ created_at : chr \u0026quot;Thu Jul 07 03:22:16 +0000 2016\u0026quot; ## $ favourites_count : int 18066 ## $ verified : logi FALSE ## $ statuses_count : int 5545 ## $ profile_image_url_https: chr \u0026quot;https://pbs.twimg.com/profile_images/1330619806396067845/mIPmR-x4_normal.jpg\u0026quot; ## $ profile_banner_url : chr \u0026quot;https://pbs.twimg.com/profile_banners/750892662224453632/1602417664\u0026quot; ## $ default_profile : logi FALSE ## $ default_profile_image : logi FALSE ## $ withheld_in_countries :List of 1 ## ..$ : list() ## $ entities :List of 1 ## ..$ :List of 2 ## - attr(*, \u0026quot;tweets\u0026quot;)='data.frame':\t1 obs. of 36 variables: ## ..$ created_at : chr \u0026quot;Thu Jul 29 16:15:52 +0000 2021\u0026quot; ## ..$ id : num 1.42e+18 ## ..$ id_str : chr \u0026quot;1420780122949529601\u0026quot; ## ..$ text : chr \u0026quot;Looking forward to digging into this paper \\U0001f929\\U0001f9ee\\U0001f60d https://t.co/ddCMraM3kj\u0026quot; ## ..$ truncated : logi FALSE ## ..$ entities :List of 1 ## ..$ source : chr \u0026quot;\u0026lt;a href=\\\u0026quot;http://twitter.com/download/iphone\\\u0026quot; rel=\\\u0026quot;nofollow\\\u0026quot;\u0026gt;Twitter for iPhone\u0026lt;/a\u0026gt;\u0026quot; ## ..$ in_reply_to_status_id : logi NA ## ..$ in_reply_to_status_id_str : logi NA ## ..$ in_reply_to_user_id : logi NA ## ..$ in_reply_to_user_id_str : logi NA ## ..$ in_reply_to_screen_name : logi NA ## ..$ geo : logi NA ## ..$ coordinates :List of 1 ## ..$ place :List of 1 ## ..$ contributors : logi NA ## ..$ is_quote_status : logi TRUE ## ..$ quoted_status_id : num 1.42e+18 ## ..$ quoted_status_id_str : chr \u0026quot;1420768369620312065\u0026quot; ## ..$ retweet_count : int 0 ## ..$ favorite_count : int 8 ## ..$ favorited : logi FALSE ## ..$ retweeted : logi FALSE ## ..$ possibly_sensitive : logi FALSE ## ..$ lang : chr \u0026quot;en\u0026quot; ## ..$ quoted_status :List of 1 ## ..$ display_text_width : int 70 ## ..$ user :List of 1 ## ..$ full_text : logi NA ## ..$ favorited_by : logi NA ## ..$ display_text_range : logi NA ## ..$ retweeted_status : logi NA ## ..$ quoted_status_permalink : logi NA ## ..$ metadata : logi NA ## ..$ query : logi NA ## ..$ possibly_sensitive_appealable: logi NA Now I obtain the id of my followers using get_followers.\nfollowers \u0026lt;- get_followers(user = \u0026quot;oaggimenez\u0026quot;, n = og$followers_count, retryonratelimit = T) From their id, I can get the same details I got on my account using lookup_users. This function is not vectorized, therefore I use a loop. Takes some time so I saved the results and load them.\ndetails_followers \u0026lt;- NULL for (i in 1:length(followers$user_id)){ tmp \u0026lt;- try(lookup_users(followers$user_id[i], retryonratelimit = TRUE), silent = TRUE) if (length(tmp) == 1){ next } else { tmp$listed_count \u0026lt;- NULL # get rid of this column which raised some format issues, we do not it anyway details_followers \u0026lt;- bind_rows(details_followers, tmp) } } save(details_followers, file = \u0026quot;details_followers.RData\u0026quot;) What info do we have?\nload(\u0026quot;dat/details_followers.RData\u0026quot;) names(details_followers) ## [1] \u0026quot;id\u0026quot; \u0026quot;id_str\u0026quot; ## [3] \u0026quot;name\u0026quot; \u0026quot;screen_name\u0026quot; ## [5] \u0026quot;location\u0026quot; \u0026quot;description\u0026quot; ## [7] \u0026quot;url\u0026quot; \u0026quot;protected\u0026quot; ## [9] \u0026quot;followers_count\u0026quot; \u0026quot;friends_count\u0026quot; ## [11] \u0026quot;created_at\u0026quot; \u0026quot;favourites_count\u0026quot; ## [13] \u0026quot;verified\u0026quot; \u0026quot;statuses_count\u0026quot; ## [15] \u0026quot;profile_image_url_https\u0026quot; \u0026quot;profile_banner_url\u0026quot; ## [17] \u0026quot;default_profile\u0026quot; \u0026quot;default_profile_image\u0026quot; ## [19] \u0026quot;withheld_in_countries\u0026quot; \u0026quot;entities\u0026quot; In more details.\nstr(details_followers, max = 1) ## 'data.frame':\t5025 obs. of 20 variables: ## $ id : num 1.03e+18 1.42e+18 1.32e+09 1.42e+18 1.36e+18 ... ## $ id_str : chr \u0026quot;1026053171024736258\u0026quot; \u0026quot;1419513064697712640\u0026quot; \u0026quot;1317282258\u0026quot; \u0026quot;1419222504367927296\u0026quot; ... ## $ name : chr \u0026quot;Dr. Zoe Nhleko\u0026quot; \u0026quot;akash Shelke\u0026quot; \u0026quot;Rilwan Ugbedeojo\u0026quot; \u0026quot;Thaana Van Dessel\u0026quot; ... ## $ screen_name : chr \u0026quot;ZoeNhleko\u0026quot; \u0026quot;akashSh88945929\u0026quot; \u0026quot;abuhrilwan\u0026quot; \u0026quot;ThaanaD\u0026quot; ... ## $ location : chr \u0026quot;United States\u0026quot; \u0026quot;\u0026quot; \u0026quot;Lagos, Nigeria\u0026quot; \u0026quot;France\u0026quot; ... ## $ description : chr \u0026quot;Wildlife ecologist with experience on African large mammals. PhD from @UF. Looking for job opportunities in eco\u0026quot;| __truncated__ \u0026quot;\u0026quot; \u0026quot;MSc, Plant Ecology | Seeking PhD Position in Community Ecology | Civic Leader | YALI RLC \u0026amp; African Presidential\u0026quot;| __truncated__ \u0026quot;MSc student majoring in Animal Ecology @WURanimal. \\nFocused on Conservation Behavior \u0026amp; Human-Wildlife Coexistence.\u0026quot; ... ## $ url : chr \u0026quot;https://t.co/uaFY2L8BnL\u0026quot; NA NA NA ... ## $ protected : logi FALSE FALSE FALSE FALSE FALSE FALSE ... ## $ followers_count : int 2878 1 244 0 2856 4 112 6 69 476 ... ## $ friends_count : int 1325 223 2030 14 2575 99 344 32 73 1671 ... ## $ created_at : chr \u0026quot;Sun Aug 05 10:31:53 +0000 2018\u0026quot; \u0026quot;Mon Jul 26 04:21:18 +0000 2021\u0026quot; \u0026quot;Sat Mar 30 22:35:27 +0000 2013\u0026quot; \u0026quot;Sun Jul 25 09:06:36 +0000 2021\u0026quot; ... ## $ favourites_count : int 8003 0 1843 1 12360 0 484 0 83 713 ... ## $ verified : logi FALSE FALSE FALSE FALSE FALSE FALSE ... ## $ statuses_count : int 7261 0 670 1 2112 0 141 0 30 926 ... ## $ profile_image_url_https: chr \u0026quot;https://pbs.twimg.com/profile_images/1306353853571280896/rk9qAxoa_normal.jpg\u0026quot; \u0026quot;https://pbs.twimg.com/profile_images/1419513134142738433/erR2XRwf_normal.png\u0026quot; \u0026quot;https://pbs.twimg.com/profile_images/1413368052012433413/EyzbYfWB_normal.jpg\u0026quot; \u0026quot;https://pbs.twimg.com/profile_images/1419222962566217731/TQT96npR_normal.jpg\u0026quot; ... ## $ profile_banner_url : chr \u0026quot;https://pbs.twimg.com/profile_banners/1026053171024736258/1533465589\u0026quot; NA \u0026quot;https://pbs.twimg.com/profile_banners/1317282258/1624183687\u0026quot; \u0026quot;https://pbs.twimg.com/profile_banners/1419222504367927296/1627205254\u0026quot; ... ## $ default_profile : logi TRUE TRUE FALSE TRUE TRUE TRUE ... ## $ default_profile_image : logi FALSE FALSE FALSE FALSE FALSE FALSE ... ## $ withheld_in_countries :List of 5025 ## $ entities :List of 5025 ## - attr(*, \u0026quot;tweets\u0026quot;)='data.frame':\t1 obs. of 36 variables: Data exploration and visualisation Let\u0026rsquo;s display the 100 bigger accounts that follow me. First thing I learned. It is humbling to be followed by influential individuals I admire like @MicrobiomDigest, @nathanpsmad, @FrancoisTaddei, @freakonometrics, @allison_horst, @HugePossum, @apreshill and scientific journals, institutions and societies.\ndetails_followers %\u0026gt;% arrange(-followers_count) %\u0026gt;% select(screen_name, followers_count, friends_count, favourites_count) %\u0026gt;% head(n = 100) ## screen_name followers_count friends_count favourites_count ## 1 dianefrancis1 230560 132635 70 ## 2 simongerman600 218444 209489 56976 ## 3 WildlifeMag 191613 23995 14697 ## 4 cagrimbakirci 170164 9401 21334 ## 5 BernardMarr 130900 29532 15764 ## 6 ZotovMax 119625 133610 894 ## 7 _atanas_ 116717 74266 102899 ## 8 MicrobiomDigest 105825 33779 98199 ## 9 Eliances 97488 94488 15164 ## 10 PLOSBiology 89263 6646 4278 ## 11 IPBES 78086 11797 53957 ## 12 7wData 76131 84155 74609 ## 13 biconnections 63802 52918 91970 ## 14 aydintufekci43 60212 22627 99427 ## 15 AliceVachet 56381 2320 182435 ## 16 Pr_Logos 52482 50520 19852 ## 17 DrJoeNyangon 51296 20858 1413 ## 18 derekantoncich 48823 51449 347 ## 19 Afro_Herper 45215 3392 117693 ## 20 KumarAGarg 43461 47795 4052 ## 21 GatelyMark 41500 37944 1153 ## 22 overleaf 41306 7830 35289 ## 23 figshare 40610 38361 10395 ## 24 mlamons1 40415 29364 2498 ## 25 coraman 40398 2169 16548 ## 26 BritishEcolSoc 37394 1143 11685 ## 27 nathanpsmad 36760 5312 43729 ## 28 WildlifeRescue 34843 7976 61284 ## 29 rudyagovic 34063 28147 3708 ## 30 grp_resilience 34011 7998 3559 ## 31 JNakev 33427 28532 29442 ## 32 FrancoisTaddei 32631 17903 23894 ## 33 RunEducator 31248 19848 17522 ## 34 LarryLLapin 30782 34265 501 ## 35 _Alex_Iaco_ 30775 26720 2390 ## 36 owasow 30562 12932 36909 ## 37 JEcology 30534 689 3312 ## 38 valmasdel 28972 4512 4334 ## 39 freakonometrics 28038 15112 31106 ## 40 MethodsEcolEvol 26207 10347 2083 ## 41 mmw_lmw 26014 12157 1068 ## 42 drmichellelarue 25424 6281 56120 ## 43 nicolasberrod 25258 6042 12990 ## 44 DD_NaNa_ 24253 11051 39916 ## 45 callin_bull 24106 2831 10572 ## 46 INEE_CNRS 23483 1873 5089 ## 47 Datascience__ 22793 10695 216 ## 48 VisualPersist 22779 5599 79872 ## 49 coywolfassoc 22203 19491 9470 ## 50 RBGSydney 20259 16049 13435 ## 51 allison_horst 19289 2865 7725 ## 52 rhskraus 18429 20029 3503 ## 53 fred_sallee 18144 18995 3423 ## 54 thembauk 17748 2066 3260 ## 55 parolesdhist 17257 7136 7497 ## 56 CREAF_ecologia 16654 3201 11899 ## 57 ChamboncelLea 16609 4059 9797 ## 58 HugePossum 16461 12732 7806 ## 59 ScientistFemale 16379 9039 2079 ## 60 sophie_e_hill 16292 5811 10315 ## 61 mohamma64508589 16130 14550 24584 ## 62 SamuelHayat 15999 3068 9767 ## 63 MathieuAvanzi 15890 1604 8526 ## 64 Erky321 15859 16284 4841 ## 65 ConLetters 15482 5547 1489 ## 66 oceansresearch 14969 1759 7599 ## 67 WCSNewsroom 14926 11224 7277 ## 68 CMastication 14687 8091 31769 ## 69 apreshill 14106 1607 32806 ## 70 SumanMalla10 13872 11977 10946 ## 71 umontpellier 13667 1360 8745 ## 72 WriteNThrive 13615 13523 7631 ## 73 abdirashidmd 13586 2389 18417 ## 74 JackKinross 13392 11263 774 ## 75 PaulREhrlich 13383 14208 1 ## 76 teppofelin 13330 12410 264 ## 77 ConservOptimism 13249 4306 15648 ## 78 jepeteso 13137 13595 3420 ## 79 gwintrob 12484 8751 43479 ## 80 SallyR_ISLHE 12411 12477 28924 ## 81 ActuIAFr 12321 8856 2495 ## 82 Booksmag 12282 1785 332 ## 83 jusoneoftoomany 12212 11049 2296 ## 84 galaxyproject 12208 8757 813 ## 85 thepostdoctoral 12186 12802 814 ## 86 FrontMarineSci 11498 5733 5832 ## 87 jobRxiv 11442 872 786 ## 88 EcographyJourna 11381 1465 4938 ## 89 davidghamilton1 11242 2231 16244 ## 90 5amStartups 11170 12370 147062 ## 91 RosmarieKatrin 11079 11159 4060 ## 92 ScienceAndMaps 11043 8679 3596 ## 93 joelgombin 11012 10335 10868 ## 94 SteveLenhert 10990 8328 669 ## 95 Naturevolve 10973 11349 7500 ## 96 cookinforpeace 10655 2861 17519 ## 97 DrPaulEvans1 10537 7927 3131 ## 98 Ibycter 10517 4450 47359 ## 99 Sciencecomptoir 10436 1268 15800 ## 100 SteveAtLogikk 10421 8864 13123 Where do my followers come from? Second thing I learnt. Followers are folks from France (my home), USA, UK, Canada, Australia, and other European countries. I was happy to realize that I have followers from other parts of the world too, in India, South Africa, Brazil, Peru, Mexico, Nepal, Argentina, Kenya, Bangladesh, etc. and from Greece (my other home).\ndetails_followers %\u0026gt;% mutate(location = str_replace(location, \u0026quot;.*France.*\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*England\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*United Kingdom\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*London\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*Scotland\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*Montpellier\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*Canada, BC.*\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*New Caledonia\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*Marseille\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*Grenoble\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*Perú\u0026quot;, \u0026quot;Peru\u0026quot;), location = str_replace(location, \u0026quot;.*Martinique\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*La Rochelle\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*Grenoble, Rhône-Alpes\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*Avignon\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*Paris\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*paris\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*france\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*Bordeaux\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*Mytilene, Lesvos, Greece\u0026quot;, \u0026quot;Greece\u0026quot;), location = str_replace(location, \u0026quot;.*Mytiline, Greece\u0026quot;, \u0026quot;Greece\u0026quot;), location = str_replace(location, \u0026quot;.*Germany.*\u0026quot;, \u0026quot;Germany\u0026quot;), location = str_replace(location, \u0026quot;.*Vienna, Austria\u0026quot;, \u0026quot;Austria\u0026quot;), location = str_replace(location, \u0026quot;.*Arusha, Tanzania\u0026quot;, \u0026quot;Tanzania\u0026quot;), location = str_replace(location, \u0026quot;.*Athens\u0026quot;, \u0026quot;Greece\u0026quot;), location = str_replace(location, \u0026quot;.*Bangor, ME\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*Bangor, Wales\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*Baton Rouge, LA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Lisboa, Portugal\u0026quot;, \u0026quot;Portugal\u0026quot;), location = str_replace(location, \u0026quot;.*Oslo\u0026quot;, \u0026quot;Norway\u0026quot;), location = str_replace(location, \u0026quot;.*Islamic Republic of Iran\u0026quot;, \u0026quot;Iran\u0026quot;), location = str_replace(location, \u0026quot;.*Madrid, España\u0026quot;, \u0026quot;Spain\u0026quot;), location = str_replace(location, \u0026quot;.*Madrid\u0026quot;, \u0026quot;Spain\u0026quot;), location = str_replace(location, \u0026quot;.*Madrid, Spain\u0026quot;, \u0026quot;Spain\u0026quot;), location = str_replace(location, \u0026quot;.*Rome, Italy\u0026quot;, \u0026quot;Italy\u0026quot;), location = str_replace(location, \u0026quot;.*Santiago, Chile\u0026quot;, \u0026quot;Chile\u0026quot;), location = str_replace(location, \u0026quot;.*Belém, Brazil\u0026quot;, \u0026quot;Brazil\u0026quot;), location = str_replace(location, \u0026quot;.*Belo Horizonte, Brazil\u0026quot;, \u0026quot;Brazil\u0026quot;), location = str_replace(location, \u0026quot;.*Berlin\u0026quot;, \u0026quot;Germany\u0026quot;), location = str_replace(location, \u0026quot;.*Mexico, ME\u0026quot;, \u0026quot;Mexico\u0026quot;), location = str_replace(location, \u0026quot;.*Mexico City\u0026quot;, \u0026quot;Mexico\u0026quot;), location = str_replace(location, \u0026quot;.*Santiago, Chile\u0026quot;, \u0026quot;Chile\u0026quot;), location = str_replace(location, \u0026quot;.*Ontario\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Aberdeen\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*Cambridge\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*Bologna, Emilia Romagna\u0026quot;, \u0026quot;Italy\u0026quot;), location = str_replace(location, \u0026quot;.*Bogotá, D.C., Colombia\u0026quot;, \u0026quot;Colombia\u0026quot;), location = str_replace(location, \u0026quot;.*Medellín, Colombia\u0026quot;, \u0026quot;Colombia\u0026quot;), location = str_replace(location, \u0026quot;.*Bruxelles, Belgique\u0026quot;, \u0026quot;Belgium\u0026quot;), location = str_replace(location, \u0026quot;.*Brasil\u0026quot;, \u0026quot;Brazil\u0026quot;), location = str_replace(location, \u0026quot;.*Budweis, Czech Republic\u0026quot;, \u0026quot;Czech Republic\u0026quot;), location = str_replace(location, \u0026quot;.*Calgary, Alberta\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Canada, BC\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Cardiff, Wales\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*Brisbane\u0026quot;, \u0026quot;Australia\u0026quot;), location = str_replace(location, \u0026quot;.*Sydney\u0026quot;, \u0026quot;Australia\u0026quot;), location = str_replace(location, \u0026quot;.*Queensland\u0026quot;, \u0026quot;Australia\u0026quot;), location = str_replace(location, \u0026quot;.*Australia\u0026quot;, \u0026quot;Australia\u0026quot;), location = str_replace(location, \u0026quot;.*Germany\u0026quot;, \u0026quot;Germany\u0026quot;), location = str_replace(location, \u0026quot;.*Vancouver\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Ottawa, Ontario\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Québec, Canada\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Winnipeg, Manitoba\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*New South Wales\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Victoria\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*British Columbia\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Norway\u0026quot;, \u0026quot;Norway\u0026quot;), location = str_replace(location, \u0026quot;.*Finland\u0026quot;, \u0026quot;Finland\u0026quot;), location = str_replace(location, \u0026quot;.*South Africa\u0026quot;, \u0026quot;South Africa\u0026quot;), location = str_replace(location, \u0026quot;.*Switzerland\u0026quot;, \u0026quot;Switzerland\u0026quot;), location = str_replace(location, \u0026quot;.*CO\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*OK\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*KS\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*MS\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*CO\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*CO\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*CO\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*CO\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*WA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*MD\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Colorado\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Community of Valencia, Spain\u0026quot;, \u0026quot;Spain\u0026quot;), location = str_replace(location, \u0026quot;.*Dunedin City, New Zealand\u0026quot;, \u0026quot;New Zealand\u0026quot;), location = str_replace(location, \u0026quot;.*Rio Claro, Brasil\u0026quot;, \u0026quot;Brazil\u0026quot;), location = str_replace(location, \u0026quot;.*Saskatoon, Saskatchewan\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Sherbrooke, Québec\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*United Kingdom.*\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*University of Oxford\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*University of St Andrews\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*ID\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*NE\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*United States of America, USA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Spain, Spain\u0026quot;, \u0026quot;Spain\u0026quot;), location = str_replace(location, \u0026quot;.*USA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Wisconsin, USA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Florida, USA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Liege, Belgium\u0026quot;, \u0026quot;Belgium\u0026quot;), location = str_replace(location, \u0026quot;.*Ghent, Belgium\u0026quot;, \u0026quot;Belgium\u0026quot;), location = str_replace(location, \u0026quot;.*Pune, India\u0026quot;, \u0026quot;India\u0026quot;), location = str_replace(location, \u0026quot;.*Hyderabad, India\u0026quot;, \u0026quot;India\u0026quot;), location = str_replace(location, \u0026quot;.*Prague, Czech Republic\u0026quot;, \u0026quot;Czech Republic\u0026quot;), location = str_replace(location, \u0026quot;.*Canada, BC, Canada\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*CA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*United States.*\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*DC\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*FL\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*GA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*HI\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*ME\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*MA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*MI\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*PA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*NC\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*MO\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*NY\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*NH\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*IL\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*NM\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*MT\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*OR\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*WY\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*WI\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*MN\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*CT\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*TX\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*VA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*OH\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Massachusetts, USA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*California, USA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Montréal, Québec\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Edmonton, Alberta\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Toronto, Ontario\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Canada, Canada\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Montreal\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Lisbon, Portugal\u0026quot;, \u0026quot;Portugal\u0026quot;), location = str_replace(location, \u0026quot;.*Coimbra, Portugal\u0026quot;, \u0026quot;Portugal\u0026quot;), location = str_replace(location, \u0026quot;.*Cork, Ireland\u0026quot;, \u0026quot;Ireland\u0026quot;), location = str_replace(location, \u0026quot;.*Dublin City, Ireland\u0026quot;, \u0026quot;Ireland\u0026quot;), location = str_replace(location, \u0026quot;.*Barcelona, Spain\u0026quot;, \u0026quot;Spain\u0026quot;), location = str_replace(location, \u0026quot;.*Barcelona\u0026quot;, \u0026quot;Spain\u0026quot;), location = str_replace(location, \u0026quot;.*Leipzig\u0026quot;, \u0026quot;Germany\u0026quot;), location = str_replace(location, \u0026quot;.*Seville, Spain\u0026quot;, \u0026quot;Spain\u0026quot;), location = str_replace(location, \u0026quot;.*Seville, Spain\u0026quot;, \u0026quot;Spain\u0026quot;), location = str_replace(location, \u0026quot;.*Buenos Aires, Argentina\u0026quot;, \u0026quot;Argentina\u0026quot;), location = str_replace(location, \u0026quot;.*Rio de Janeiro, Brazil\u0026quot;, \u0026quot;Brazil\u0026quot;), location = str_replace(location, \u0026quot;.*Canberra\u0026quot;, \u0026quot;Australia\u0026quot;), location = str_remove(location, \u0026quot;Global\u0026quot;), location = str_remove(location, \u0026quot;Earth\u0026quot;), location = str_remove(location, \u0026quot;Worldwide\u0026quot;), location = str_remove(location, \u0026quot;Europe\u0026quot;), location = str_remove(location, \u0026quot; \u0026quot;), location = str_replace(location, \u0026quot;.*Dhaka, Bangladesh\u0026quot;, \u0026quot;Bangladesh\u0026quot;), location = str_replace(location, \u0026quot;.*Copenhagen, Denmark\u0026quot;, \u0026quot;Denmark\u0026quot;), location = str_replace(location, \u0026quot;.*Amsterdam, The Netherlands\u0026quot;, \u0026quot;The Netherlands\u0026quot;), location = str_replace(location, \u0026quot;.*Groningen, Nederland\u0026quot;, \u0026quot;The Netherlands\u0026quot;), location = str_replace(location, \u0026quot;.*Wageningen, Nederland\u0026quot;, \u0026quot;The Netherlands\u0026quot;), location = str_replace(location, \u0026quot;.*Aarhus, Denmark\u0026quot;, \u0026quot;Denmark\u0026quot;), location = str_replace(location, \u0026quot;.*Antwerp, Belgium\u0026quot;, \u0026quot;Belgium\u0026quot;), location = str_replace(location, \u0026quot;.*Aveiro, Portugal\u0026quot;, \u0026quot;Portugal\u0026quot;), location = str_replace(location, \u0026quot;.*Australia, AUS\u0026quot;, \u0026quot;Australia\u0026quot;), location = str_replace(location, \u0026quot;.*Australian National University\u0026quot;, \u0026quot;Australia\u0026quot;), location = str_replace(location, \u0026quot;.*Auckland, New Zealand\u0026quot;, \u0026quot;New Zealand\u0026quot;), location = str_replace(location, \u0026quot;.*Belfast, Northern Ireland\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*Ireland\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*Hobart, Tasmania\u0026quot;, \u0026quot;Australia\u0026quot;), location = str_replace(location, \u0026quot;.*Dhaka, Bangladesh\u0026quot;, \u0026quot;Bangladesh\u0026quot;), location = str_replace(location, \u0026quot;.*Nairobi, Kenya\u0026quot;, \u0026quot;Kenya\u0026quot;), location = str_replace(location, \u0026quot;.*Dhaka, Bangladesh\u0026quot;, \u0026quot;Bangladesh\u0026quot;), location = str_replace(location, \u0026quot;.*Berlin, Deutschland\u0026quot;, \u0026quot;Germany\u0026quot;), location = str_replace(location, \u0026quot;.*Munich, Bavaria\u0026quot;, \u0026quot;Germany\u0026quot;), location = str_replace(location, \u0026quot;.*Dehradun, India\u0026quot;, \u0026quot;India\u0026quot;), location = str_replace(location, \u0026quot;.*Bengaluru, India\u0026quot;, \u0026quot;India\u0026quot;), location = str_replace(location, \u0026quot;.*Berlin, Deutschland\u0026quot;, \u0026quot;Germany\u0026quot;), location = str_replace(location, \u0026quot;.*Deutschland\u0026quot;, \u0026quot;Germany\u0026quot;), location = str_replace(location, \u0026quot;.*Edinburgh\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*Glasgow\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*New York, USA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Washington, USA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*California\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*California\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Christchurch City, New Zealand\u0026quot;, \u0026quot;New Zealand\u0026quot;), location = str_replace(location, \u0026quot;.*Harare, Zimbabwe\u0026quot;, \u0026quot;Zimbabwe\u0026quot;), location = str_replace(location, \u0026quot;.*Islamabad, Pakistan\u0026quot;, \u0026quot;Pakistan\u0026quot;), location = str_replace(location, \u0026quot;.*Kolkata, India\u0026quot;, \u0026quot;India\u0026quot;), location = str_replace(location, \u0026quot;.*Lagos, Nigeria\u0026quot;, \u0026quot;Nigeria\u0026quot;), location = str_replace(location, \u0026quot;.*Lima, Peru\u0026quot;, \u0026quot;Peru\u0026quot;), location = str_replace(location, \u0026quot;.*Valparaíso, Chile\u0026quot;, \u0026quot;Chile\u0026quot;), location = str_replace(location, \u0026quot;.*Uppsala, Sweden\u0026quot;, \u0026quot;Sweden\u0026quot;), location = str_replace(location, \u0026quot;.*Uppsala, Sverige\u0026quot;, \u0026quot;Sweden\u0026quot;), location = str_replace(location, \u0026quot;.*Stockholm, Sweden\u0026quot;, \u0026quot;Sweden\u0026quot;), location = str_replace(location, \u0026quot;.*Stockholm\u0026quot;, \u0026quot;Sweden\u0026quot;), location = str_replace(location, \u0026quot;.*Turin, Piedmont\u0026quot;, \u0026quot;Italy\u0026quot;), location = str_replace(location, \u0026quot;.*University of Iceland\u0026quot;, \u0026quot;Iceland\u0026quot;), location = str_replace(location, \u0026quot;.*University of Helsinki\u0026quot;, \u0026quot;Finland\u0026quot;), location = str_replace(location, \u0026quot;.*Tucumán, Argentina\u0026quot;, \u0026quot;Argentina\u0026quot;), location = str_replace(location, \u0026quot;.*The Hague, The Netherlands\u0026quot;, \u0026quot;The Netherlands\u0026quot;), location = str_replace(location, \u0026quot;.*UK\u0026quot;, \u0026quot;United Kingdom\u0026quot;) ) %\u0026gt;% count(location, sort = TRUE) %\u0026gt;% #slice(-40) %\u0026gt;% head(n = 45) -\u0026gt; followers_by_country followers_by_country ## location n ## 1 1273 ## 2 United States of America 606 ## 3 France 571 ## 4 United Kingdom 473 ## 5 Canada 166 ## 6 Germany 130 ## 7 Australia 112 ## 8 Norway 50 ## 9 Switzerland 43 ## 10 Spain 34 ## 11 India 33 ## 12 South Africa 32 ## 13 Brazil 30 ## 14 Portugal 27 ## 15 Finland 26 ## 16 Sweden 18 ## 17 Belgium 17 ## 18 Italy 16 ## 19 The Netherlands 14 ## 20 New Zealand 13 ## 21 Peru 13 ## 22 Denmark 11 ## 23 Mexico 11 ## 24 Nepal 10 ## 25 Argentina 9 ## 26 Kenya 9 ## 27 Bangladesh 8 ## 28 Chile 8 ## 29 Greece 8 ## 30 Colombia 6 ## 31 Czech Republic 6 ## 32 Singapore 6 ## 33 Iceland 5 ## 34 Senegal 5 ## 35 Iran 4 ## 36 Nigeria 4 ## 37 Pakistan 4 ## 38 Tanzania 4 ## 39 Zimbabwe 4 ## 40 3 ## 41 Austria 3 ## 42 Ecuador 3 ## 43 Guatemala 3 ## 44 Hong Kong 3 ## 45 Luxembourg 3 Let\u0026rsquo;s have a look to the same info on a quick and dirty map. White countries do not appear in the 45 countries with most followers, or do not have any followers.\nlibrary(rworldmap) library(classInt) spdf \u0026lt;- joinCountryData2Map(followers_by_country, joinCode=\u0026quot;NAME\u0026quot;, nameJoinColumn=\u0026quot;location\u0026quot;, verbose=TRUE) ## 43 codes from your data successfully matched countries in the map ## 2 codes from your data failed to match with a country code in the map ## failedCodes failedCountries ## [1,] \u0026quot;\u0026quot; \u0026quot;\u0026quot; ## [2,] \u0026quot;\u0026quot; \u0026quot; \u0026quot; ## 200 codes from the map weren't represented in your data classInt \u0026lt;- classIntervals(spdf$n, n=9, style = \u0026quot;jenks\u0026quot;) catMethod \u0026lt;- classInt[[\u0026quot;brks\u0026quot;]] library(RColorBrewer) colourPalette \u0026lt;- brewer.pal(9,'RdPu') mapParams \u0026lt;- mapCountryData(spdf, nameColumnToPlot=\u0026quot;n\u0026quot;, addLegend=FALSE, catMethod = catMethod, colourPalette=colourPalette, mapTitle=\u0026quot;Number of followers per country\u0026quot;) do.call(addMapLegend, c(mapParams, legendLabels=\u0026quot;all\u0026quot;, legendWidth=0.5, legendIntervals=\u0026quot;data\u0026quot;, legendMar = 2)) How many followers are verified?\ndetails_followers %\u0026gt;% count(verified) ## verified n ## 1 FALSE 5009 ## 2 TRUE 16 Who are the verified users? Scientific entities, but also some scientists.\ndetails_followers %\u0026gt;% filter(verified==TRUE) %\u0026gt;% select(name, screen_name, location, followers_count) %\u0026gt;% arrange(-followers_count) ## name screen_name ## 1 Çağrı Mert Bakırcı cagrimbakirci ## 2 Elisabeth Bik MicrobiomDigest ## 3 ipbes IPBES ## 4 Dr. Earyn McGee, Lizard lassoer \\U0001f98e Afro_Herper ## 5 figshare figshare ## 6 Nathan Peiffer-Smadja nathanpsmad ## 7 Wildlife Alliance WildlifeRescue ## 8 Arthur Charpentier freakonometrics ## 9 L'écologie au CNRS INEE_CNRS ## 10 The Royal Botanic Garden Sydney RBGSydney ## 11 CREAF CREAF_ecologia ## 12 Jean-Paul Carrera JeanCarrPaul ## 13 Deckset decksetapp ## 14 𝙳𝚞𝚌-𝚀𝚞𝚊𝚗𝚐 𝙽𝚐𝚞𝚢𝚎𝚗 duc_qn ## 15 Christophe Josset \\U0001f4dd CJosset ## 16 Casey Fung TheCaseyFung ## location followers_count ## 1 Texas, USA 170164 ## 2 San Francisco, CA 105825 ## 3 Bonn, Germany 78086 ## 4 United States 45215 ## 5 London 40610 ## 6 Paris - London 36760 ## 7 Phnom Penh, Cambodia 34843 ## 8 Montréal, Québec 28038 ## 9 23483 ## 10 Sydney, New South Wales 20259 ## 11 Bellaterra (Barcelona) 16654 ## 12 Oxford, England 5744 ## 13 Your Mac 3884 ## 14 Lausanne, Switzerland 2808 ## 15 Paris, France 2002 ## 16 Bundjalung Country | Mullum 1612 How old are my followers, where age is defined as the time elapsed since they created their Twitter account. Twitter was created in 2006. I created my account in July 2016 but started using it only two or three years ago I\u0026rsquo;d say. This dataviz is not too much informative I guess.\ndetails_followers %\u0026gt;% mutate(month = created_at %\u0026gt;% str_split(\u0026quot; \u0026quot;) %\u0026gt;% map_chr(2), day = created_at %\u0026gt;% str_split(\u0026quot; \u0026quot;) %\u0026gt;% map_chr(3), year = created_at %\u0026gt;% str_split(\u0026quot; \u0026quot;) %\u0026gt;% map_chr(6), created_at = paste0(month,\u0026quot;-\u0026quot;,year), created_at = lubridate::my(created_at)) %\u0026gt;% group_by(created_at) %\u0026gt;% summarise(counts = n()) %\u0026gt;% ggplot(aes(x = created_at, y = counts)) + geom_line() + labs(x = \u0026quot;date of creation\u0026quot;, y = \u0026quot;how many accounts were created this month-year\u0026quot;) What is the distribution of the number of followers and friends. On average, 400 followers and 400 friends.\nfoll \u0026lt;- details_followers %\u0026gt;% ggplot() + aes(x = log(followers_count)) + geom_histogram(fill=\u0026quot;#69b3a2\u0026quot;, color=\u0026quot;#e9ecef\u0026quot;, alpha=0.9) + labs(x = \u0026quot;log number of followers\u0026quot;, y = \u0026quot;counts\u0026quot;, title = \u0026quot;followers\u0026quot;) friends \u0026lt;- details_followers %\u0026gt;% ggplot() + aes(x = log(friends_count)) + geom_histogram(fill=\u0026quot;#69b3a2\u0026quot;, color=\u0026quot;#e9ecef\u0026quot;, alpha=0.9) + labs(x = \u0026quot;log number of friends\u0026quot;, y = \u0026quot;\u0026quot;, title = \u0026quot;friends\u0026quot;) library(patchwork) foll + friends Retrieving social links To investigate the social network made of my followers, we need to gather who they follow, and their followers. Unfortunately, the Twitter API imposes some time and quantity constraints in retrieving data.\nHere, for the sake of illustration, I will focus on some accounts only. To do so, I consider accounts with between 1000 and 1500 followers, and more than 30000 tweets users have liked in their account\u0026rsquo;s lifetime (which denotes some activity). The more followers we consider, the more time it would take to gather info on them: it is approx one minute per follower on average, so one hour for sixty followers, and more than eighty hours for all my followers! With the filters I\u0026rsquo;m using, I end up with 35 followers here, this is very few to say infer meaningful about a network, but it\u0026rsquo;ll do the job for now.\nsome_followers \u0026lt;- details_followers %\u0026gt;% filter((followers_count \u0026gt; 1000 \u0026amp; followers_count \u0026lt; 1500), favourites_count \u0026gt; 30000) # %\u0026gt;% nrow() Create empty list and name it after their screen name.\nfoler \u0026lt;- vector(mode = 'list', length = length(some_followers$screen_name)) names(foler) \u0026lt;- some_followers$screen_name Get followers of these selected followers. Takes ages, so save and load later.\nfor (i in 1:length(some_followers$screen_name)) { message(\u0026quot;Getting followers for user #\u0026quot;, i, \u0026quot;/\u0026quot;, nrow(some_followers)) foler[[i]] \u0026lt;- get_followers(some_followers$screen_name[i], n = some_followers$followers_count[i], retryonratelimit = TRUE) if(i %% 5 == 0){ message(\u0026quot;sleep for 5 minutes\u0026quot;) Sys.sleep(5*60) } } save(foler, file = \u0026quot;foler.RData\u0026quot;) Format the data.\nload(\u0026quot;dat/foler.RData\u0026quot;) folerx \u0026lt;- bind_rows(foler, .id = \u0026quot;screen_name\u0026quot;) active_fol_x \u0026lt;- some_followers %\u0026gt;% select(id_str, screen_name) foler_join \u0026lt;- left_join(folerx, some_followers, by = \u0026quot;screen_name\u0026quot;) algo_follower \u0026lt;- foler_join %\u0026gt;% select(id_str, screen_name) %\u0026gt;% setNames(c(\u0026quot;follower\u0026quot;, \u0026quot;active_user\u0026quot;)) %\u0026gt;% na.omit() Get friends of my followers. Takes ages. Again I save and load later.\nfriend \u0026lt;- data.frame() for (i in seq_along(some_followers$screen_name)) { message(\u0026quot;Getting following for user #\u0026quot;, i ,\u0026quot;/\u0026quot;,nrow(some_followers)) kk \u0026lt;- get_friends(some_followers$screen_name[i], n = some_followers$friends_count[i], retryonratelimit = TRUE) friend \u0026lt;- rbind(friend, kk) if(i %% 15 == 0){ message(\u0026quot;sleep for 15 minutes\u0026quot;) Sys.sleep(15*60+1) } } save(friend, file = \u0026quot;friend.RData\u0026quot;) Format the data.\nload(\u0026quot;dat/friend.RData\u0026quot;) all_friend \u0026lt;- friend %\u0026gt;% setNames(c(\u0026quot;screen_name\u0026quot;, \u0026quot;user_id\u0026quot;)) all_friendx \u0026lt;- left_join(all_friend, active_fol_x, by=\u0026quot;screen_name\u0026quot;) algo_friend \u0026lt;- all_friendx %\u0026gt;% select(user_id, screen_name) %\u0026gt;% setNames(c(\u0026quot;following\u0026quot;,\u0026quot;active_user\u0026quot;)) Now that we have all info on followers and friends, we\u0026rsquo;re gonna build the network of people who follow each other.\nun_active \u0026lt;- unique(algo_friend$active_user) %\u0026gt;% data.frame(stringsAsFactors = F) %\u0026gt;% setNames(\u0026quot;active_user\u0026quot;) algo_mutual \u0026lt;- data.frame() for (i in seq_along(un_active$active_user)){ aa \u0026lt;- algo_friend %\u0026gt;% filter(active_user == un_active$active_user[i]) bb \u0026lt;- aa %\u0026gt;% filter(aa$following %in% algo_follower$follower) %\u0026gt;% setNames(c(\u0026quot;mutual\u0026quot;,\u0026quot;active_user\u0026quot;)) algo_mutual \u0026lt;- rbind(algo_mutual,bb) } Instead of ids, we use screen names.\ndetail_friend \u0026lt;- lookup_users(algo_mutual$mutual) algo_mutual \u0026lt;- algo_mutual %\u0026gt;% left_join(detail_friend, by = c(\u0026quot;mutual\u0026quot; = \u0026quot;id_str\u0026quot;)) %\u0026gt;% na.omit() %\u0026gt;% select(mutual, active_user, screen_name) algo_mutual ## # A tibble: 18 × 3 ## mutual active_user screen_name ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; ## 1 824265743575306243 g33k5p34k robbie_emmet ## 2 761735501707567104 g33k5p34k LauraBlissEco ## 3 1439272374 alexnicolharper LeafyEricScott ## 4 824265743575306243 alexnicolharper robbie_emmet ## 5 21467726 alexnicolharper g33k5p34k ## 6 4181949437 LegalizeBrain EvpokPadding ## 7 824265743575306243 Souzam7139 robbie_emmet ## 8 888890201673580548 Souzam7139 mellenmartin ## 9 824265743575306243 mellenmartin robbie_emmet ## 10 21467726 LauraBlissEco g33k5p34k ## 11 831398755920445440 JosiahParry RoelandtN42 ## 12 4181949437 gau EvpokPadding ## 13 21467726 robbie_emmet g33k5p34k ## 14 4159201575 robbie_emmet alexnicolharper ## 15 888890201673580548 robbie_emmet mellenmartin ## 16 29538964 EvpokPadding gau ## 17 1439272374 lifedispersing LeafyEricScott ## 18 301687349 RoelandtN42 JosiahParry Add my account to the network.\nun_active \u0026lt;- un_active %\u0026gt;% mutate(mutual = rep(\u0026quot;oaggimenez\u0026quot;)) un_active \u0026lt;- un_active[,c(2,1)] un_active \u0026lt;- un_active %\u0026gt;% setNames(c(\u0026quot;active_user\u0026quot;,\u0026quot;screen_name\u0026quot;)) algo_mutual \u0026lt;- bind_rows(algo_mutual %\u0026gt;% select(-mutual), un_active) For what follows, we will need packages to work with networks.\nlibrary(igraph) library(tidygraph) library(ggraph) Now we create the edges, nodes and build the network.\nnodes \u0026lt;- data.frame(V = unique(c(algo_mutual$screen_name,algo_mutual$active_user)), stringsAsFactors = F) edges \u0026lt;- algo_mutual %\u0026gt;% setNames(c(\u0026quot;from\u0026quot;,\u0026quot;to\u0026quot;)) network_ego1 \u0026lt;- graph_from_data_frame(d = edges, vertices = nodes, directed = F) %\u0026gt;% as_tbl_graph() Network metrics Create communities using group_louvain() algorithm, and calculate standard metrics using tidygraph.\nset.seed(123) network_ego1 \u0026lt;- network_ego1 %\u0026gt;% activate(nodes) %\u0026gt;% mutate(community = as.factor(group_louvain())) %\u0026gt;% mutate(degree_c = centrality_degree()) %\u0026gt;% mutate(betweenness_c = centrality_betweenness(directed = F,normalized = T)) %\u0026gt;% mutate(closeness_c = centrality_closeness(normalized = T)) %\u0026gt;% mutate(eigen = centrality_eigen(directed = F)) network_ego_df \u0026lt;- as.data.frame(network_ego1) Identify key nodes with respect to network metrics.\nnetwork_ego_df ## name community degree_c betweenness_c closeness_c eigen ## 1 robbie_emmet 2 8 0.0049299720 0.5384615 0.5238676 ## 2 LauraBlissEco 2 3 0.0000000000 0.5147059 0.2717415 ## 3 LeafyEricScott 5 3 0.0008403361 0.5223881 0.2287709 ## 4 g33k5p34k 2 6 0.0024649860 0.5303030 0.4342056 ## 5 EvpokPadding 3 4 0.0011204482 0.5223881 0.2336052 ## 6 mellenmartin 2 4 0.0000000000 0.5223881 0.3371889 ## 7 RoelandtN42 4 3 0.0000000000 0.5147059 0.2050991 ## 8 alexnicolharper 2 5 0.0019607843 0.5303030 0.3942456 ## 9 gau 3 3 0.0000000000 0.5147059 0.2133909 ## 10 JosiahParry 4 3 0.0000000000 0.5147059 0.2050991 ## 11 genius_c137 1 1 0.0000000000 0.5072464 0.1454399 ## 12 Zjbb 1 1 0.0000000000 0.5072464 0.1454399 ## 13 SJRAfloat 1 1 0.0000000000 0.5072464 0.1454399 ## 14 BMPARMA17622540 1 1 0.0000000000 0.5072464 0.1454399 ## 15 maryam_adeli 1 1 0.0000000000 0.5072464 0.1454399 ## 16 waywardaf 1 1 0.0000000000 0.5072464 0.1454399 ## 17 SusyVF6 1 1 0.0000000000 0.5072464 0.1454399 ## 18 Sinalo_NM 1 1 0.0000000000 0.5072464 0.1454399 ## 19 Leila_Lula 1 1 0.0000000000 0.5072464 0.1454399 ## 20 CARThorpe 1 1 0.0000000000 0.5072464 0.1454399 ## 21 j_wilson_white 1 1 0.0000000000 0.5072464 0.1454399 ## 22 LegalizeBrain 3 2 0.0000000000 0.5147059 0.1794154 ## 23 JaishriJuice 1 1 0.0000000000 0.5072464 0.1454399 ## 24 jaguaretepy 1 1 0.0000000000 0.5072464 0.1454399 ## 25 Souzam7139 2 3 0.0000000000 0.5223881 0.2706719 ## 26 LaLince_ 1 1 0.0000000000 0.5072464 0.1454399 ## 27 Nina_Ella_ 1 1 0.0000000000 0.5072464 0.1454399 ## 28 samcox 1 1 0.0000000000 0.5072464 0.1454399 ## 29 TAdamsBio42 1 1 0.0000000000 0.5072464 0.1454399 ## 30 brubakerl 1 1 0.0000000000 0.5072464 0.1454399 ## 31 robanhk 1 1 0.0000000000 0.5072464 0.1454399 ## 32 InesCCarv 1 1 0.0000000000 0.5072464 0.1454399 ## 33 groundhog0202 1 1 0.0000000000 0.5072464 0.1454399 ## 34 zenmart 1 1 0.0000000000 0.5072464 0.1454399 ## 35 lifedispersing 5 2 0.0000000000 0.5147059 0.1787123 ## 36 oaggimenez 1 35 0.9685154062 1.0000000 1.0000000 Get users with highest values of each metric.\nkp_ego \u0026lt;- data.frame( network_ego_df %\u0026gt;% arrange(-degree_c) %\u0026gt;% select(name), network_ego_df %\u0026gt;% arrange(-betweenness_c) %\u0026gt;% select(name), network_ego_df %\u0026gt;% arrange(-closeness_c) %\u0026gt;% select(name), network_ego_df %\u0026gt;% arrange(-eigen) %\u0026gt;% select(name)) %\u0026gt;% setNames(c(\u0026quot;degree\u0026quot;,\u0026quot;betweenness\u0026quot;,\u0026quot;closeness\u0026quot;,\u0026quot;eigen\u0026quot;)) kp_ego[-1,] ## degree betweenness closeness eigen ## 2 robbie_emmet robbie_emmet robbie_emmet robbie_emmet ## 3 g33k5p34k g33k5p34k g33k5p34k g33k5p34k ## 4 alexnicolharper alexnicolharper alexnicolharper alexnicolharper ## 5 EvpokPadding EvpokPadding LeafyEricScott mellenmartin ## 6 mellenmartin LeafyEricScott EvpokPadding LauraBlissEco ## 7 LauraBlissEco LauraBlissEco mellenmartin Souzam7139 ## 8 LeafyEricScott mellenmartin Souzam7139 EvpokPadding ## 9 RoelandtN42 RoelandtN42 LauraBlissEco LeafyEricScott ## 10 gau gau RoelandtN42 gau ## 11 JosiahParry JosiahParry gau RoelandtN42 ## 12 Souzam7139 genius_c137 JosiahParry JosiahParry ## 13 LegalizeBrain Zjbb LegalizeBrain LegalizeBrain ## 14 lifedispersing SJRAfloat lifedispersing lifedispersing ## 15 genius_c137 BMPARMA17622540 genius_c137 robanhk ## 16 Zjbb maryam_adeli Zjbb Sinalo_NM ## 17 SJRAfloat waywardaf SJRAfloat samcox ## 18 BMPARMA17622540 SusyVF6 BMPARMA17622540 genius_c137 ## 19 maryam_adeli Sinalo_NM maryam_adeli Zjbb ## 20 waywardaf Leila_Lula waywardaf SJRAfloat ## 21 SusyVF6 CARThorpe SusyVF6 BMPARMA17622540 ## 22 Sinalo_NM j_wilson_white Sinalo_NM waywardaf ## 23 Leila_Lula LegalizeBrain Leila_Lula SusyVF6 ## 24 CARThorpe JaishriJuice CARThorpe Leila_Lula ## 25 j_wilson_white jaguaretepy j_wilson_white j_wilson_white ## 26 JaishriJuice Souzam7139 JaishriJuice JaishriJuice ## 27 jaguaretepy LaLince_ jaguaretepy jaguaretepy ## 28 LaLince_ Nina_Ella_ LaLince_ LaLince_ ## 29 Nina_Ella_ samcox Nina_Ella_ Nina_Ella_ ## 30 samcox TAdamsBio42 samcox brubakerl ## 31 TAdamsBio42 brubakerl TAdamsBio42 groundhog0202 ## 32 brubakerl robanhk brubakerl zenmart ## 33 robanhk InesCCarv robanhk maryam_adeli ## 34 InesCCarv groundhog0202 InesCCarv CARThorpe ## 35 groundhog0202 zenmart groundhog0202 InesCCarv ## 36 zenmart lifedispersing zenmart TAdamsBio42 Robbie Emmet has the highest degree, betweenness, closeness centrality and eigenvector centrality. He has the most relations with the other nodes in the network. He also can spread information the further away and faster than anyone, and is also surrounded by important persons in the network. This is within the network we\u0026rsquo;ve just built.\nBy the way, Robbie has just passed his dissertation defense, congrats! Follow him on Twitter at @robbie_emmet, he is awesome!\nVisualize Network We visualize the network using clusters or communities (first three only).\noptions(ggrepel.max.overlaps = Inf) network_viz \u0026lt;- network_ego1 %\u0026gt;% filter(community %in% 1:3) %\u0026gt;% mutate(node_size = ifelse(degree_c \u0026gt;= 50,degree_c,0)) %\u0026gt;% mutate(node_label = ifelse(betweenness_c \u0026gt;= 0.01,name,NA)) plot_ego \u0026lt;- network_viz %\u0026gt;% ggraph(layout = \u0026quot;stress\u0026quot;) + geom_edge_fan(alpha = 0.05) + geom_node_point(aes(color = as.factor(community),size = node_size)) + geom_node_label(aes(label = node_label),nudge_y = 0.1, show.legend = F, fontface = \u0026quot;bold\u0026quot;, fill = \u0026quot;#ffffff66\u0026quot;) + theme_graph() + theme(legend.position = \u0026quot;none\u0026quot;) + labs(title = \u0026quot;Top 3 communities\u0026quot;) plot_ego Let\u0026rsquo;s use an alternative dataviz to identify the nodes.\nnetwork_viz %\u0026gt;% mutate(community = group_spinglass()) %\u0026gt;% ggraph(layout = \u0026quot;nicely\u0026quot;) + geom_edge_fan(alpha = 0.25) + geom_node_point(aes(color = factor(community)),size = 5, show.legend = F) + geom_node_text(aes(label = name),repel = T) + theme_graph() + theme(legend.position = \u0026quot;none\u0026quot;) + labs(title = \u0026quot;Network with named nodes\u0026quot;) Concluding words Not sure what the communities are about. Due to limitation with the Twitter API I did not attempt to retrieve the info on all the 5,000 followers, but I used only 35 of them, which explains why there is not much to say about this network I guess. I might take some time to try and retrieve all the info (will take days), or not. I might also use some information I already have, like how followers describe themselves and do some topic modelling to identify common themes of interest.\nHere I have simply followed the steps that Joe Cristian illustrated so brillantly in his post at https://algotech.netlify.app/blog/social-network-analysis-in-r/. Make sure you read his post for more details on the code and theory. In particular, he illustrates how information spreads through the network.\nFor more about social network analyses and Twitter, see this thread https://twitter.com/Mehdi_Moussaid/status/1389174715990876160?s=20 and this video https://www.youtube.com/watch?v=UX7YQ6m2r_o in French, this is what inspired me to do the analysis. See also this https://github.com/eleurent/twitter-graph for a much better analysis than what I could do (with Python for data retrieving and manipulation, and Gephi for network visualisation).\nThis post was also published on https://www.r-bloggers.com.\u0026#160;\u0026#x21a9;\u0026#xfe0e;\n","date":1627430400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1627430400,"objectID":"40f652687c71880ce5f02374b6de0deb","permalink":"https://oliviergimenez.github.io/blog/twitter-social-network/","publishdate":"2021-07-28T00:00:00Z","relpermalink":"/blog/twitter-social-network/","section":"blog","summary":"I use R to retrieve some data from Twitter, do some exploratory data analysis and visualisation and examine a network of followers.\n","tags":["Twitter","social network","R","rstats"],"title":"Quick and dirty analysis of a Twitter social network","type":"blog"},{"authors":null,"categories":null,"content":"It was awesome!\nBig day today 🥳 Hélène Labach is defending her PhD on the conservation of Bottlenose 🐬 in the French Mediterranean 🌊 @MIRACETI_ceta @cefemontpellier @umontpellier @CollegeDoctoUM @twitthair1 pic.twitter.com/RkbHSe415Q\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 24, 2021 ","date":1624492800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1624492800,"objectID":"052faa6ce386aa5343660bba1182d1a7","permalink":"https://oliviergimenez.github.io/blog/lnphddefense/","publishdate":"2021-06-24T00:00:00Z","relpermalink":"/blog/lnphddefense/","section":"blog","summary":"It was awesome!\n","tags":["population ecology","Conservation biology"],"title":"Hélène Labach's PhD defense on the conservation of Bottlenose dolphins in the French Mediterranean sea","type":"blog"},{"authors":null,"categories":null,"content":"Slides and codes available from https://sarahbauduin.github.io/formation_IBM_NetLogoR/.\nOur workshop on individual-based models #IBMs with NetLogoR #rstats starts today, led by #twitterless IBM expert Sarah Bauduin @OFBiodiversite @cefemontpellier w/ @oksanagrente \u0026amp; @NSantostasi\n▶️ Slides \u0026amp; codes https://t.co/LjCh4AH0UA\n▶️ NetLogoR https://t.co/KXQxIGwKMh pic.twitter.com/MjvVZxMi7H\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 21, 2021 ","date":1624320000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1624320000,"objectID":"2c5acfbe92fe4bfe53b206237f9d55a8","permalink":"https://oliviergimenez.github.io/blog/ibmworkshop/","publishdate":"2021-06-22T00:00:00Z","relpermalink":"/blog/ibmworkshop/","section":"blog","summary":"Slides and codes available from https://sarahbauduin.github.io/formation_IBM_NetLogoR/.\n","tags":["rstats","individual-based models"],"title":"Workshop on individual-based models with R","type":"blog"},{"authors":null,"categories":null,"content":"Our poster with Ruth King and Rachel McCrea for the 2021 Euring conference.\nPoster session at @_EURING #euring2021 feels like a video game from back in the days @gather_town 🥳 Sadly, couldn\u0026#39;t stay as I was on 🧇 duty 😋 Btw our poster w/ @BayesRuth \u0026amp; @RSMcCrea is a review of the capture-recapture literature over the last decade, check out ⤵️ pic.twitter.com/B2Gal2Sj6I\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 8, 2021 ","date":1623110400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1623110400,"objectID":"c0985876325312cf08f9afcc1bad8262","permalink":"https://oliviergimenez.github.io/blog/euring2021/","publishdate":"2021-06-08T00:00:00Z","relpermalink":"/blog/euring2021/","section":"blog","summary":"Our poster with Ruth King and Rachel McCrea for the 2021 Euring conference.\n","tags":["Capture-recapture","Literature review","Conference","Poster"],"title":"A review of the capture-recapture literature over the last decade","type":"blog"},{"authors":null,"categories":null,"content":"Journée de restitution sur notre projet ERC-Lynx.\nLancement de la journée de restitution de notre projet #ittecop \u0026quot;Eviter, réduire et compenser la mortalité du Lynx par collision avec les véhicules de transport\u0026quot; @ademe @FRBiodiv @Ecologie_Gouv Plus sur le projet ici https://t.co/uKY5ylOli2 pic.twitter.com/e0TdSqvhUZ\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 4, 2021 ","date":1622764800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1622764800,"objectID":"5912b711a934ee57ac239c8bc5277898","permalink":"https://oliviergimenez.github.io/blog/restitutionerc/","publishdate":"2021-06-04T00:00:00Z","relpermalink":"/blog/restitutionerc/","section":"blog","summary":"Journée de restitution sur notre projet ERC-Lynx.\n","tags":["Lynx lynx","Car-wildlife collisions","Human-Wildlife conflicts"],"title":"Journée de restitution du projet ERC Lynx","type":"blog"},{"authors":null,"categories":null,"content":"Everything (including data, slides, codes and video recordings) about the workshop we run on Bayesian capture-recapture inference with hidden Markov models in R and Nimble.\n🥳🍻 Had fun running a workshop on Bayesian capture-recapture inference with hidden Markov models in R #rstats \u0026amp; Nimble w/ @SarahCubaynes @chloe_nater @MaudQueroue \u0026amp; Perry de Valpine @R_nimble ✅ Slides, worksheets \u0026amp; video recordings freely available from https://t.co/CgVd84MPYN pic.twitter.com/WlHzXytVUG\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) May 18, 2021 ","date":1621296000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1621296000,"objectID":"7f3925442bafb377695f4d37d73b3dac","permalink":"https://oliviergimenez.github.io/blog/hmmcrworkshop/","publishdate":"2021-05-18T00:00:00Z","relpermalink":"/blog/hmmcrworkshop/","section":"blog","summary":"Everything (including data, slides, codes and video recordings) about the workshop we run on Bayesian capture-recapture inference with hidden Markov models in R and Nimble.\n","tags":["Bayesian statistics","Capture-recapture models","rstats","Nimble"],"title":"Workshop on Bayesian capture-recapture inference","type":"blog"},{"authors":null,"categories":null,"content":"Everything (including data, slides and codes) about the workshop on Bayesian statistics I run.\nJust finished up a week teaching Bayesian statistics with R #rstats for non-stats PhD students \u0026amp; 💜 it #bayes #mcmc #jags\n🌍 Dedicated website https://t.co/X7JY3f9eGX\n🍳 Material https://t.co/agY9CvQLiV\n📽️ Video recording (in English w/ 👽 accent) https://t.co/DHabXmVYsT pic.twitter.com/g94nSO3faQ\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 26, 2021 ","date":1616716800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1616716800,"objectID":"e7a5270e0c95180e8190f0abce1d80d6","permalink":"https://oliviergimenez.github.io/blog/bayesstatsworkshop/","publishdate":"2021-03-26T00:00:00Z","relpermalink":"/blog/bayesstatsworkshop/","section":"blog","summary":"Everything (including data, slides and codes) about the workshop on Bayesian statistics I run.\n","tags":["Bayesian statistics","rstats"],"title":"Workshop on Bayesian statistics with R","type":"blog"},{"authors":null,"categories":null,"content":"Everything (including data, slides and codes) about the workshop on reproducible science I run.\nHad fun running a 1-day workshop on reproducible science in R \u0026amp; RStudio for our lab @cefemontpellier. We covered data manip \u0026amp; visualisation w/ #tidyverse, version control w/ Git/GitHub and writing dynamic docs w/ R Markdown\n📝 Slides \u0026amp; practicals https://t.co/LJRknr8Kzb 📽️=🇫🇷 pic.twitter.com/uCsm9kH0Hf\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 21, 2021 ","date":1616284800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1616284800,"objectID":"53cfcbee95df7f8b1c2d6d03f2858f60","permalink":"https://oliviergimenez.github.io/blog/rrworkshop/","publishdate":"2021-03-21T00:00:00Z","relpermalink":"/blog/rrworkshop/","section":"blog","summary":"Everything (including data, slides and codes) about the workshop on reproducible science I run.\n","tags":["Reproducible science","rstats","RStudio"],"title":"Workshop on reproducible science in R \u0026 RStudio","type":"blog"},{"authors":null,"categories":null,"content":"More about our project on mitigating collisions between lynx and cars.\n📽️🎙️ La vidéo et les diapos du webinaire sont disponibles https://t.co/xt2EXsvIGN - 🙏 aux participant.e.s pour leurs questions, et à l\u0026#39;équipe #Ittecop pour l\u0026#39;organisation ! https://t.co/Ix7Zkse02H\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 8, 2021 ","date":1615161600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1615161600,"objectID":"01a1a74013cd423dcfb3aa2f0678a497","permalink":"https://oliviergimenez.github.io/blog/commerc/","publishdate":"2021-03-08T00:00:00Z","relpermalink":"/blog/commerc/","section":"blog","summary":"More about our project on mitigating collisions between lynx and cars.\n","tags":["Human-wildlife conflicts","Lynx lynx","Car-wildlife collisions","SciComm"],"title":"Communication sur notre projet ERC-Lynx","type":"blog"},{"authors":null,"categories":null,"content":"Our take on deep learning, computer vision and ecology.\n\u0026quot;Images, écologie et deep learning\u0026quot; co-écrit avec Vincent Miele et Stéphane Dray @LbbeLyon sur les applications du #DeepLearning à l\u0026#39;analyse d\u0026#39;images en écologie et difficultés associées. 🙏 @sfecologie pour l\u0026#39;opportunité https://t.co/0cTYbHTb35 @INEE_CNRS @cefemontpellier\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 22, 2021 ","date":1613952000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1613952000,"objectID":"59d0320da40df81624533f20d2410842","permalink":"https://oliviergimenez.github.io/blog/sfeeia/","publishdate":"2021-02-22T00:00:00Z","relpermalink":"/blog/sfeeia/","section":"blog","summary":"Our take on deep learning, computer vision and ecology.\n","tags":["Artificial intelligence","Deep learning","Computer vision","GDR EcoStat"],"title":"Regard SFEE Images, écologie et deep learning","type":"blog"},{"authors":null,"categories":null,"content":"Some highlights on the year 2020 for our group.\nOur research group had a fantastic 2020 year! A thread w/ some highlights 🥳🍾 https://t.co/BuNSAceisM 0/9 pic.twitter.com/uPuAjV5uve\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 23, 2020 ","date":1608768000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1608768000,"objectID":"b93a279462a5c9f64e26813b903746b6","permalink":"https://oliviergimenez.github.io/blog/retrospective2020/","publishdate":"2020-12-24T00:00:00Z","relpermalink":"/blog/retrospective2020/","section":"blog","summary":"Some highlights on the year 2020 for our group.\n","tags":["laughter","retrospective","research group"],"title":"We had a great 2020 year, considering...","type":"blog"},{"authors":null,"categories":null,"content":"Workshop to come on reproducible science in our lab.\n🧑‍🔬💻♻️ Planning a 1-day workshop on reproducible science for our lab @cefemontpellier. We\u0026#39;ll use R and RStudio, Git, GitHub, R Markdown, tidyverse and sf 🤩\nMaterial here https://t.co/QApKQsl1FU #rstats #ReproducibleScience @twitthair1 pix by @scriberian pic.twitter.com/LFBwZmelit\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 18, 2020 ","date":1608249600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1608249600,"objectID":"10792b7b86f34b3580cdb7e37ae7eee8","permalink":"https://oliviergimenez.github.io/blog/reproscience2020/","publishdate":"2020-12-18T00:00:00Z","relpermalink":"/blog/reproscience2020/","section":"blog","summary":"Workshop to come on reproducible science in our lab.\n","tags":["reproducible science","R","rstat","git","GitHub","tidyverse"],"title":"Workshop on reproducible science","type":"blog"},{"authors":null,"categories":null,"content":"A seminar I gave remotely for the university of Lisboa.\nStatistical ecology 🌍📊, large carnivores 🐺🐻 and a pinch of social sciences 🤷, happy to share our research in this 45\u0026#39; talk 🤩🍾\nslides: https://t.co/fKTMWko5J9\nvideo: https://t.co/7KubqYJIBU pic.twitter.com/OJ292VWuel\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 10, 2020 ","date":1607558400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1607558400,"objectID":"f3bc928c9dd62acad9b683a7f3a6a5c8","permalink":"https://oliviergimenez.github.io/blog/talklisboa/","publishdate":"2020-12-10T00:00:00Z","relpermalink":"/blog/talklisboa/","section":"blog","summary":"A seminar I gave remotely for the university of Lisboa.\n","tags":["large carnivores","statistical ecology","seminar"],"title":"Seminar on large carnivores and statistical ecology","type":"blog"},{"authors":null,"categories":null,"content":"Just finished a workshop on integrated population models.\nAnd it\u0026#39;s oooooooooooon ! https://t.co/sEy70RCORN pic.twitter.com/nNBCvIonjV\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) November 23, 2020 ","date":1606089600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1606089600,"objectID":"a5701df93c48a04007f484b65fa379e7","permalink":"https://oliviergimenez.github.io/blog/ipmworkshop_thread/","publishdate":"2020-11-23T00:00:00Z","relpermalink":"/blog/ipmworkshop_thread/","section":"blog","summary":"Just finished a workshop on integrated population models.\n","tags":["integrated population models","capture-recapture","workshop"],"title":"Workshop on integrated population models","type":"blog"},{"authors":null,"categories":null,"content":"Co-organisé les journées annuelles Mate-shs pendant lesquelles on a parlé des nouveaux métiers et outils autour des données en SHS.\nLes journées annuelles Mate-shs se tiennent aujourd\u0026#39;hui et demain 🎉🤩 On y parle des nouveaux métiers et outils autour des données en SHS @mate_shs_cnrs @MSHsud @CNRS @INSHS_CNRS Site web : https://t.co/oDJsi9deUo\nProgramme : https://t.co/kKkklKhEsd#SHS #interdisciplinarité pic.twitter.com/bliBnZm8qY\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) November 19, 2020 ","date":1605744000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1605744000,"objectID":"c525b990816a0ac4f3cee8ab5016fd3d","permalink":"https://oliviergimenez.github.io/blog/mateshs/","publishdate":"2020-11-19T00:00:00Z","relpermalink":"/blog/mateshs/","section":"blog","summary":"Co-organisé les journées annuelles Mate-shs pendant lesquelles on a parlé des nouveaux métiers et outils autour des données en SHS.\n","tags":["SHS","science des données","sciences sociales","journées"],"title":"Journées annuelles réseau Mate-SHS","type":"blog"},{"authors":null,"categories":null,"content":"Co-organised a workshop on deep learning applications in ecology. Thread below and my notes there.\n🥁🤩 Our workshop on #DeepLearning and #Ecology is about to begin. We\u0026#39;ll be talking about how to process and analyse images 📸 and sounds 🎙️ Program: https://t.co/gUXD0tMCW1\nCurated list of resources for ecologists: https://t.co/HieQfqgVEx GDR EcoStat: https://t.co/kTXpjJXlya pic.twitter.com/nfjVke9ETU\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) November 16, 2020 ","date":1605484800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1605484800,"objectID":"469f2c2df4be9c6aa3893dd54ca1b03a","permalink":"https://oliviergimenez.github.io/blog/deeplearning/","publishdate":"2020-11-16T00:00:00Z","relpermalink":"/blog/deeplearning/","section":"blog","summary":"Co-organised a workshop on deep learning applications in ecology. Thread below and my notes there.\n","tags":["deep learning","workshops","computer vision","statistical ecology"],"title":"Workshop on Deep Learning and Ecology","type":"blog"},{"authors":null,"categories":null,"content":"The material (slides and video) of my course on statistics for ecologists for Master students.\nEvery year I teach a 12h course on statistics for ecologists for Master students at @umontpellier covering freq/bayes GLM(M)s. The material of the course is available ➡️ https://t.co/Y1CTvkmH8Q w/ slides, practicals \u0026amp; videos (in French).#StatisticalEcology #ReproducibleTeaching pic.twitter.com/A15vndiw49\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) November 12, 2020 ","date":1605139200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1605139200,"objectID":"a207a18f3cc4f797b6446e2e54202ce1","permalink":"https://oliviergimenez.github.io/blog/stat-course-2020/","publishdate":"2020-11-12T00:00:00Z","relpermalink":"/blog/stat-course-2020/","section":"blog","summary":"The material (slides and video) of my course on statistics for ecologists for Master students.\n","tags":["statistical ecology","Master course"],"title":"Master course on statistics for ecologists","type":"blog"},{"authors":null,"categories":null,"content":"Notre article pour la Fête de la Science dans The Conversation \u0026ldquo; Mieux connaître le lynx boréal grâce à l\u0026rsquo;écologie statistique\u0026rdquo;.\nJetez un coup d\u0026#39;oeil à notre article pour la Fête de la Science dans @FR_Conversation \u0026quot;Mieux connaître le lynx boréal grâce à l\u0026#39;écologie statistique\u0026quot; https://t.co/d40TzNcmis co-écrit avec @LouvrierJulie \u0026amp; Sarah Bauduin - avec l\u0026#39;aide précieuse de Françoise Marmouyet #FDS2020 pic.twitter.com/xmJX0crmmy\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) October 8, 2020 ","date":1602115200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1602115200,"objectID":"da6e1b0a4adaea0943753d6a9fd5ea4a","permalink":"https://oliviergimenez.github.io/blog/theconversation/","publishdate":"2020-10-08T00:00:00Z","relpermalink":"/blog/theconversation/","section":"blog","summary":"Notre article pour la Fête de la Science dans The Conversation \u0026ldquo; Mieux connaître le lynx boréal grâce à l\u0026rsquo;écologie statistique\u0026rdquo;.\n","tags":["The Conversation","écologie statistique","lynx boréal","scicomm"],"title":"Article The Conversation sur le lynx et l'écologie statistique","type":"blog"},{"authors":null,"categories":null,"content":"Gathered some code on occupancy, capture-recapture \u0026amp; epidemiological models, social networks, spatial stuff, textual analyses, reproducible science, etc\u0026hellip;\n👩‍💻👨‍💻 In hope it\u0026#39;s useful, I gathered random pieces of code I wrote on occupancy, capture-recapture \u0026amp; epidemiological models, social networks, spatial stuff, textual analyses, #DeepLearning and the #tidyverse ➡️ https://t.co/T1PlbmDLZO I also went full purple @MoonbeamLevels 💜 pic.twitter.com/q6vK4LYVS7\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) August 7, 2020 ","date":1596758400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1596758400,"objectID":"d19a6bfefff5762e590b101a802dcb2d","permalink":"https://oliviergimenez.github.io/blog/pieceofcode/","publishdate":"2020-08-07T00:00:00Z","relpermalink":"/blog/pieceofcode/","section":"blog","summary":"Gathered some code on occupancy, capture-recapture \u0026amp; epidemiological models, social networks, spatial stuff, textual analyses, reproducible science, etc\u0026hellip;\n","tags":["R","Nimble","Jags","coding"],"title":"Some random piece of code","type":"blog"},{"authors":null,"categories":["project"],"content":"Large carnivores are difficult to monitor in the field: they are elusive and live at low densities over wide areas. With their recent recovery in Europe, large carnivores are having increasing interactions with human activities. We contribute to the estimation of their abundance and distribution by applying and developing methods that make optimal use of available data. We are also working on applying science-based management using the structured decision making framework. Past and ongoing projects were are are mostly led by Master students, PhDs and post-docs.\n","date":1596672000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1596672000,"objectID":"3d31a390c54fc2c129512091d2cf0878","permalink":"https://oliviergimenez.github.io/my-project/external-project/carnivores/","publishdate":"2020-08-06T00:00:00Z","relpermalink":"/my-project/external-project/carnivores/","section":"my-project","summary":"Developing statistical tools to survey large carnivores in Europe.","tags":["carnivores","monitoring","management"],"title":"Large carnivores monitoring","type":"my-project"},{"authors":null,"categories":null,"content":"","date":1596672000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1596672000,"objectID":"dc306a4b5877ae8802b6579f50075fd7","permalink":"https://oliviergimenez.github.io/my-project/external-project/lynx/","publishdate":"2020-08-06T00:00:00Z","relpermalink":"/my-project/external-project/lynx/","section":"my-project","summary":"Avoiding, reducing and compensating mortality risks of the Eurasian Lynx by collision with transport vehicles.","tags":["carnivores","collisions","management","software"],"title":"Terrestrial transport infrastructures, Lynx population viability and mitigation strategies","type":"my-project"},{"authors":null,"categories":null,"content":"","date":1595203200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1595203200,"objectID":"077e1e01fd8590c86aff39607410dbcb","permalink":"https://oliviergimenez.github.io/project_landing/","publishdate":"2020-07-20T00:00:00Z","relpermalink":"/project_landing/","section":"","summary":"See some of the projects I have worked on.","tags":null,"title":"Projects","type":"widget_page"},{"authors":null,"categories":null,"content":"Trained in mathematics and statistics in particular, I have developed a bottomless taste for questions in statistical ecology.\nI often resort to state-space models, and hidden Markov models (e.g., here and here) in particular, to develop statistical methods for capture-recapture, occupancy and integrated population models.\nIn our group, we address questions in conservation biology to assess species viability, in ecology to study the effect of climate change on animal demography, in evolution to examine life-history tactics and individual heterogeneity and in wildlife management to develop strategies for conservation and conflict mitigation. To account for the human dimension inherent to these questions, I went back to the university to study sociology. A nice illustration of this interdisciplinary work is Gilles Maurer\u0026rsquo;s PhD combining anthropology, demography, economics and genetics to better understand the interactions between captive and wild population of Asian elephants.\nI\u0026rsquo;m doing my best to teach statistics (including the Bayesian way) to the students of the Ecology and Evolution Master in Montpellier. We also make efforts to structure the community in France through our national research group in statistical ecology. Last but not least, every now and then, our group runs workshops to diffuse quantitative methods in the ecological community.\n","date":1595203200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1595203200,"objectID":"363851c9ba77641b8792e2edf5dc8a5c","permalink":"https://oliviergimenez.github.io/my-project/external-project/methods/","publishdate":"2020-07-20T00:00:00Z","relpermalink":"/my-project/external-project/methods/","section":"my-project","summary":"Developing, communicating and teaching methods and models for ecology and conservation biology.","tags":["methods","capture-recapture","R","Bayes","MCMC"],"title":"Statistical ecology","type":"my-project"},{"authors":null,"categories":null,"content":"Just finished polishing up my slides for a workshop on Bayesian statistics.\nJust finished polishing up my slides for a workshop on Bayesian statistics 🥱😅 ☑️ Slides, code \u0026amp; data https://t.co/1SrGkYO8o0\n☑️ Videos 🎦 soon to follow (in French, then English w/ strong accent).#rstats #jags #BayesianStatistics #MCMC #EcologicalStatistics #MultilevelModels pic.twitter.com/lxleExWjqE\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 18, 2020 ","date":1595030400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1595030400,"objectID":"3ecb62d28f23fbd957c234767fac2809","permalink":"https://oliviergimenez.github.io/blog/bayesworkshop/","publishdate":"2020-07-18T00:00:00Z","relpermalink":"/blog/bayesworkshop/","section":"blog","summary":"Just finished polishing up my slides for a workshop on Bayesian statistics.\n","tags":["Bayes","workshops","statistical ecology"],"title":"Workshop on Bayesian statistics","type":"blog"},{"authors":null,"categories":null,"content":"Quick and dirty text mining of the ISEC2020 abstracts.\nQuick and dirty text mining of the @isec2020 #vISEC2020 abstracts using awesome \u0026#39;Text Mining with R - A Tidy Approach\u0026#39; by @juliasilge \u0026amp; @drob https://t.co/qwugbZevDY - codes \u0026amp; data ➡️ https://t.co/Y4adeWUpiW pic.twitter.com/bPjAgNwsWL\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 5, 220 ","date":1593993600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1593993600,"objectID":"8c08a9ad386aedfbe20564764c16e7f7","permalink":"https://oliviergimenez.github.io/blog/textminingisec/","publishdate":"2020-07-06T00:00:00Z","relpermalink":"/blog/textminingisec/","section":"blog","summary":"Quick and dirty text mining of the ISEC2020 abstracts.\n","tags":["text mining","tidyverse","isec2020"],"title":"Text mining of the ISEC2020 abstracts","type":"blog"},{"authors":null,"categories":null,"content":"Tribute to Byron Morgan\nWas visiting your website @RSMcCrea and read again the Byronmetrics paper 😍🤩 https://t.co/9XDRWagU4p that you put together back for @ISECStatEcol 2014 in Montpellier to honour Byron Morgan\u0026#39;s retirement - Thx for doing that, and say hello to Byron when you see him 😉 pic.twitter.com/E4ADQqQrBR\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 4, 220 ","date":1593820800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1593820800,"objectID":"9711fc587ab89ccb9887510f1a1a3bd5","permalink":"https://oliviergimenez.github.io/blog/byron/","publishdate":"2020-07-04T00:00:00Z","relpermalink":"/blog/byron/","section":"blog","summary":"Tribute to Byron Morgan\n","tags":["ISEC2014","Byron Morgan"],"title":"Byronmetrics","type":"blog"},{"authors":null,"categories":null,"content":"Talking about statistical ecology and large carnivores.\nTalking about statistical ecology 🌎💻📊 \u0026amp; large carnivores 🐻🐺😺 is always fun 🥳 Check out some work we\u0026#39;re doing in our @twitthair1 group https://t.co/EPZpr0FP9I, w/ guests stars 🤩@SCubi25 @LMarescot #SarahBauduin @isec2020 @ISECStatEcol @LouvrierJulie @oksanagrente pic.twitter.com/euH0TdO3Hy\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 2, 220 ","date":1591056000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1591056000,"objectID":"4c6c85d86a49203d142f31d104d932bb","permalink":"https://oliviergimenez.github.io/blog/publiclecture2/","publishdate":"2020-06-02T00:00:00Z","relpermalink":"/blog/publiclecture2/","section":"blog","summary":"Talking about statistical ecology and large carnivores.\n","tags":["large carnivores","public lecture","statistical ecology"],"title":"Statistical ecology and large carnivores","type":"blog"},{"authors":null,"categories":null,"content":"Tagging species with deep learning.\nTagging species w/ #Keras #Retinanet https://t.co/lZRlrDHtfM @FizyrBV Not too bad so far 🤩 green box is truth, blue/orange box is prediction #DeepLearning #CameraTraps More details ➡️ https://t.co/HieQfqgVEx pic.twitter.com/uhocKDuuGP\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) May 11, 220 ","date":1589155200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1589155200,"objectID":"173f1417293edf177fb74743fc4dc7a1","permalink":"https://oliviergimenez.github.io/blog/dlcameras/","publishdate":"2020-05-11T00:00:00Z","relpermalink":"/blog/dlcameras/","section":"blog","summary":"Tagging species with deep learning.\n","tags":["deep learning","camera traps"],"title":"Tagging species with deep learning.","type":"blog"},{"authors":null,"categories":null,"content":"Crude attempt at fitting dynamic ODE-BASED models to noisy data with Nimble.\nCrude attempt at fitting dynamic ODE-BASED models to noisy data with #nimble @R_nimble https://t.co/xmmPtSUR2R I got to call R functions in the code, change MCMC algorithms, do block sampling, \u0026amp; all in #rstats environment 🤩 #StatisticalEcology Thx D Turek \u0026amp; P de Valpine for help pic.twitter.com/MgyI1c5WLH\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 2020 ","date":1587686400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1587686400,"objectID":"42ac37c99f133ae8729f092ba099ff1e","permalink":"https://oliviergimenez.github.io/blog/odenimble/","publishdate":"2020-04-24T00:00:00Z","relpermalink":"/blog/odenimble/","section":"blog","summary":"Crude attempt at fitting dynamic ODE-BASED models to noisy data with Nimble.\n","tags":["ODE","Nimble"],"title":"Fitting ODE-based model to data in Nimble","type":"blog"},{"authors":null,"categories":null,"content":"Interactive data visualisation of bias in occupancy models w/ flexdashboard.\nInteractive #dataviz of bias in occupancy models w/ #flexdashboard (aka #shinyapps for the rest of us) https://t.co/fkYiHfAaEU #rstats @rstudio #unmarked https://t.co/KKGTPE8eJh pic.twitter.com/Rstpdv8CqJ\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 27,020 ","date":1587686400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1587686400,"objectID":"050895ce54c9472ccd011ae1e7c2b7ff","permalink":"https://oliviergimenez.github.io/blog/interactiveocc/","publishdate":"2020-04-24T00:00:00Z","relpermalink":"/blog/interactiveocc/","section":"blog","summary":"Interactive data visualisation of bias in occupancy models w/ flexdashboard.\n","tags":["occupancy","flexdashboard","r","shiny"],"title":"Interactive visualisation of bias in occupancy models","type":"blog"},{"authors":null,"categories":null,"content":"How to simulate data from the Rota\u0026rsquo;s multispecies occupancy model.\nI\u0026#39;m working a lot w/ multispecies occupancy models these days, needed to simulate data from models à la Rota et al. (2016), in case it is useful to others ➡️ #rstats code https://t.co/qVzaXLoYcV Feedbacks welcome 😉 See also comments ⬇️ pic.twitter.com/fDNBp48GXT\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 13, 2020 A *Bayesian* follow-up on fitting a 2-species occupancy model w/ #unmarked to #rstats simulated data: How to fit the simplest Rota\u0026#39;s model in #Jags and @R_nimble ➡️ https://t.co/BmEAx0cRvw Feedback is welcome 😊 https://t.co/PCO8pGLtLB pic.twitter.com/OpSskddTxP\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 14, 2020 ","date":1586736000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1586736000,"objectID":"b8dc6ae70c9198b95237fad1d5ba4f4d","permalink":"https://oliviergimenez.github.io/blog/rota/","publishdate":"2020-04-13T00:00:00Z","relpermalink":"/blog/rota/","section":"blog","summary":"How to simulate data from the Rota\u0026rsquo;s multispecies occupancy model.\n","tags":["occupancy","simulations","Rota"],"title":"Simulations from the Rota's multispecies occupancy model","type":"blog"},{"authors":null,"categories":null,"content":"Quick \u0026amp; dirty simulations to look at bias in occupancy estimates with respect to species detection for a static/single-season model.\nQuick \u0026amp; dirty simulations to look at bias in occupancy estimates with respect to species detection for a static/single-season model https://t.co/RRNpXuB3la #rstats #occupancy #unmarked pic.twitter.com/eN5kkdbkfm\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 12, 2020 ","date":1586649600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1586649600,"objectID":"8d1dff32845e5edf25a293bee40fe2cb","permalink":"https://oliviergimenez.github.io/blog/occusimu/","publishdate":"2020-04-12T00:00:00Z","relpermalink":"/blog/occusimu/","section":"blog","summary":"Quick \u0026amp; dirty simulations to look at bias in occupancy estimates with respect to species detection for a static/single-season model.\n","tags":["occupancy","simulations","bias"],"title":"Bias in occupancy models","type":"blog"},{"authors":null,"categories":null,"content":"Une analyse descriptive du phénomène de concentration d’attaques de loup sur des élevages d’animaux domestiques en France.\nAvec des collègues de @INRAE_France, @Supagro, #CERPAM et @OFBiodiversite, nous proposons une analyse descriptive du phénomène de concentration d’attaques de loup sur des élevages d’animaux domestiques en France https://t.co/Ueyq5gkhi1 - #rstats ➡️ https://t.co/RKtvqjBVk0 pic.twitter.com/qoup5jKauv\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 12, 2020 ","date":1586649600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1586649600,"objectID":"8efbac66c95f22a42496bd51987754e2","permalink":"https://oliviergimenez.github.io/blog/coadapht/","publishdate":"2020-04-12T00:00:00Z","relpermalink":"/blog/coadapht/","section":"blog","summary":"Une analyse descriptive du phénomène de concentration d’attaques de loup sur des élevages d’animaux domestiques en France.\n","tags":["wolf","tidyverse","large carnivores","predation"],"title":"Déprédations par le loup en France","type":"blog"},{"authors":null,"categories":null,"content":"Bayesian implementation of Pollock\u0026rsquo;s robust design capture-recapture models w/ Jags.\nBayesian implementation of Pollock\u0026#39;s @kpollock48 robust design capture-recapture models w/ #jags #rstats from a @MethodsEcolEvol paper by T. Riecke and colleagues ➡️ https://t.co/nE7N8kHPiR pic.twitter.com/ZToHY3luZR\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 6, 2020 ","date":1586131200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1586131200,"objectID":"dd2e9e8057e997ee48e695485b355129","permalink":"https://oliviergimenez.github.io/blog/rdcmr/","publishdate":"2020-04-06T00:00:00Z","relpermalink":"/blog/rdcmr/","section":"blog","summary":"Bayesian implementation of Pollock\u0026rsquo;s robust design capture-recapture models w/ Jags.\n","tags":["robust design","rstats","capture-recapture"],"title":"Bayesian implementation of the robust design","type":"blog"},{"authors":null,"categories":null,"content":"In our weekly group meeting this morning, I introduced R Markdown a great #rstats tool to write reproducible documents (reports, articles, slides, websites) smoothly mixing text, code, figures and equations in html, word or pdf format.\nIn our weekly @twitthair1 meeting this morning, I introduced R Markdown https://t.co/Hza76oAw4K a great #rstats tool to write reproducible documents (reports, articles, slides, websites) smoothly mixing text, code, figures and equations in html, word or pdf format. Ressources ⬇️ pic.twitter.com/r8vKbEaz04\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 31, 2020\u003c\u003e ","date":1585612800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1585612800,"objectID":"b597fb39065631e32071005001e61178","permalink":"https://oliviergimenez.github.io/blog/rmarkdown/","publishdate":"2020-03-31T00:00:00Z","relpermalink":"/blog/rmarkdown/","section":"blog","summary":"In our weekly group meeting this morning, I introduced R Markdown a great #rstats tool to write reproducible documents (reports, articles, slides, websites) smoothly mixing text, code, figures and equations in html, word or pdf format.","tags":["R Markdown","rstats","reproducible science"],"title":"R Markdown","type":"blog"},{"authors":null,"categories":null,"content":"In a recent seminar, S. Gandon (w/ S. Lion and T. Day) used SIR models to illustrate the effect of diff strategies of #SocialDistancing on the #COVID19 epidemic. What better way to learn than to reproduce their results in R? My two cents #rstats code ➡️ https://t.co/oJVQHJ6HqO 🤓 pic.twitter.com/dqSIcjbhqq\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 29, 2/a\u003e ","date":1585526400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1585526400,"objectID":"cd0613bb9d7b45e8b078285f4a5754c5","permalink":"https://oliviergimenez.github.io/blog/sirstats/","publishdate":"2020-03-30T00:00:00Z","relpermalink":"/blog/sirstats/","section":"blog","summary":"In a recent seminar, S. Gandon (w/ S. Lion and T. Day) used SIR models to illustrate the effect of diff strategies of #SocialDistancing on the #COVID19 epidemic. What better way to learn than to reproduce their results in R?","tags":["SIR","rstats","ODEs"],"title":"SIR models in R","type":"blog"},{"authors":null,"categories":null,"content":"Ongoing work on a flexible spatio-temporal model for occupancy data w/ a GAM-like approach in Nimble.\n😺📉 Ongoing work on a flexible spatio-temporal model for #occupancy data w/ GAM-like approach in @R_nimble\nThis work wouldn\u0026#39;t be possible wo/ the dedication of @OFBiodiversite colleagues 🙏💯 Slides➡️https://t.co/MSjwmtCCSq GAM anim courtesy @ucfagls\nThx @Yvan2935 for the tweet https://t.co/zsyTyh1Qru\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 11, 2/a\u003e ","date":1583884800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1583884800,"objectID":"ead821b85a3abaa1f765069cb57d7ffd","permalink":"https://oliviergimenez.github.io/blog/gamoccupancy/","publishdate":"2020-03-11T00:00:00Z","relpermalink":"/blog/gamoccupancy/","section":"blog","summary":"Ongoing work on a flexible spatio-temporal model for occupancy data w/ a GAM-like approach in Nimble.\n","tags":["occupancy","GAM","large carnivores"],"title":"GAM-like modelling of lynx occupancy in France","type":"blog"},{"authors":null,"categories":null,"content":"On my way to Rennes for our annual meeting on Statistical Ecology, looking forward to catching up w/ the French crowd.\nOn my way to @agrocampusouest for our annual meeting on #StatisticalEcology, looking forward to catching up w/ the french crowd #SemaineDesMaths #gdrecostat @Marie_Etienne @twitthair1 pic.twitter.com/2TGDer4wea\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 9, 2/a\u003e ","date":1583712000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1583712000,"objectID":"3bef4e6acdedeb0e8b5c8a22d013bc0b","permalink":"https://oliviergimenez.github.io/blog/gdrecostat/","publishdate":"2020-03-09T00:00:00Z","relpermalink":"/blog/gdrecostat/","section":"blog","summary":"On my way to Rennes for our annual meeting on Statistical Ecology, looking forward to catching up w/ the French crowd.\n","tags":["statistical ecology","Rennes"],"title":"French annual meeting in statistical ecology","type":"blog"},{"authors":null,"categories":null,"content":"It was so great to have you @ZipkinLab w/ us, it was too short though, come back soon 😋🚀 Monarch 🦋 are fascinating, data integration is the way to go! @twitthair1 pic.twitter.com/9ud1ycPZvE\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 6, 2/a\u003e ","date":1583452800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1583452800,"objectID":"ee66bd8c8a46af1501de1ec7122fa74a","permalink":"https://oliviergimenez.github.io/blog/elisevisit/","publishdate":"2020-03-06T00:00:00Z","relpermalink":"/blog/elisevisit/","section":"blog","summary":"It was so great to have you @ZipkinLab w/ us, it was too short though, come back soon 😋🚀 Monarch 🦋 are fascinating, data integration is the way to go! @twitthair1 pic.","tags":["community ecology","IPM"],"title":"Elise Zipkin's visit","type":"blog"},{"authors":null,"categories":null,"content":"It is my pleasure to announce that Michael Schaub and Marc Kéry will run their Integrated Population Modelling workshop in Montpellier this November.\nIt\u0026#39;s my pleasure to announce that #MichaelSchaub \u0026amp; #MarcKéry will run their Integrated Population Modelling #IPM workshop in Montpellier 🇫🇷 this November, with help from @MaudQueroue and I - details ➡️ https://t.co/pgCSxwwJW3. Come over, it\u0026#39;s going to be aaaaawesome 😁 pic.twitter.com/JvoXk07Jb1\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 4020 ","date":1583280000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1583280000,"objectID":"0f9717e88e30dc86693b47b56439cf55","permalink":"https://oliviergimenez.github.io/blog/ipmworkshop/","publishdate":"2020-03-04T00:00:00Z","relpermalink":"/blog/ipmworkshop/","section":"blog","summary":"It is my pleasure to announce that Michael Schaub and Marc Kéry will run their Integrated Population Modelling workshop in Montpellier this November.\n","tags":["IPM","workshop","rstats"],"title":"IPM workshop","type":"blog"},{"authors":null,"categories":null,"content":"Le 5 mars, l\u0026rsquo;université et la recherche s\u0026rsquo;arrêtent.\n✊ 5 mars, l\u0026#39;université et la recherche s\u0026#39;arrêtent - on est prêt 😍 #lppr #FacsEtLabosEnLutte #emploisESR Pour en savoir plus, écoutez ce #podcast https://t.co/ALGnXbgWPw @parolesdhist et si vous voulez expliquer à vos collègues 🇬🇧🇺🇸 https://t.co/iZXBsWlgFl @MHoussayH pic.twitter.com/xxKuDF6f1e\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 4, 2020 ","date":1583280000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1583280000,"objectID":"ad06ecfd93306ed1169fe829b6483f76","permalink":"https://oliviergimenez.github.io/blog/manif/","publishdate":"2020-03-04T00:00:00Z","relpermalink":"/blog/manif/","section":"blog","summary":"Le 5 mars, l\u0026rsquo;université et la recherche s\u0026rsquo;arrêtent.\n","tags":["ESR","demonstration"],"title":"L'université et la recherche s'arrêtent","type":"blog"},{"authors":null,"categories":null,"content":"I will soon be talking to an audience of Master students giving thoughts about how ecology and statistics may interact (aka Statistical Ecology), with large carnivors as a case study.\nI will soon be talking to an audience of Master students giving thoughts about how ecology \u0026amp; statistics may interact (aka #StatisticalEcology), w/ #LargeCarnivores as a case study - slides ➡️ https://t.co/ekCskObNaG. Thx @Marie_Etienne for the invit. Comments more than welcome! pic.twitter.com/A0lMPhJ5r1\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 1020 ","date":1583020800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1583020800,"objectID":"b254379c86700875ea1fc51973b674ac","permalink":"https://oliviergimenez.github.io/blog/publiclecture/","publishdate":"2020-03-01T00:00:00Z","relpermalink":"/blog/publiclecture/","section":"blog","summary":"I will soon be talking to an audience of Master students giving thoughts about how ecology and statistics may interact (aka Statistical Ecology), with large carnivors as a case study.\n","tags":["public lecture","GdR EcoStat","statistical ecology"],"title":"Public lecture on statistical ecology","type":"blog"},{"authors":null,"categories":null,"content":"Our new preprint on HMMs for ecology, or how we\u0026rsquo;'\u0026rsquo;ve been using HMMs for ever without even knowing it.\n😎😉 Our new #preprint on #HMM for ecology, or how we\u0026#39;ve been using #HMMs for ever wo even knowing it - all credit goes to @bt_mcclintock and @RolandLangrock for the heavy lifting. Written in @overleaf 😉 @EcoHMM @twitthair1 https://t.co/vSAJF5abH7\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 26020 ","date":1582675200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1582675200,"objectID":"fadf0458ceaaf1fed1e5db37ec594f94","permalink":"https://oliviergimenez.github.io/blog/reviewhmm/","publishdate":"2020-02-26T00:00:00Z","relpermalink":"/blog/reviewhmm/","section":"blog","summary":"Our new preprint on HMMs for ecology, or how we\u0026rsquo;'\u0026rsquo;ve been using HMMs for ever without even knowing it.\n","tags":["HMM","review"],"title":"Review of HMMs in ecology","type":"blog"},{"authors":null,"categories":null,"content":"Landing in Norway to meet w/ Bischof\u0026#39;s group @andyroyle_pwrc @R_nimble and brainstorm about large carnivores and spatially-explicit capture-recapture models #SCR @TouraniMa @PierreDupont47 @Milleret_Cyril the 20 degrees difference w Montpellier is... vivifying 😎🤧 pic.twitter.com/sy0ZUCoae7\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 26, 2020 WildMap happy hour with c/w from top left @eMoqanaki @TouraniMa Daniel Turek @PDupont17 @soumendey_tweet @oaggimenez R. Bischof @Milleret_Cyril Henrik Broseth, yours truly, Jonas Kindberg, @R_nimble De Valpine talking huge-scale monitoring of carnivores, SCR, Nimble, #priceybeer pic.twitter.com/AQqmTeI0Aa\n\u0026mdash; andy 😬 royle (@andyroyle_pwrc) February 26, 2020 😜 We\u0026#39;ve been busy 🐻🐺🚀 https://t.co/4npmtfZLnj pic.twitter.com/WAxqiWrDgu\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 28, 2020 ","date":1582675200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1582675200,"objectID":"23ea63008f851f49160ff2d71dee9162","permalink":"https://oliviergimenez.github.io/blog/norwaywildmap/","publishdate":"2020-02-26T00:00:00Z","relpermalink":"/blog/norwaywildmap/","section":"blog","summary":"Landing in Norway to meet w/ Bischof\u0026#39;s group @andyroyle_pwrc @R_nimble and brainstorm about large carnivores and spatially-explicit capture-recapture models #SCR @TouraniMa @PierreDupont47 @Milleret_Cyril the 20 degrees difference w Montpellier is... vivifying 😎🤧 pic.","tags":["SCR models","large carnivores","Norway"],"title":"Visit to Norway","type":"blog"},{"authors":null,"categories":null,"content":"🇮🇹🇫🇷 @NSantostasi succesfully defended her PhD on modelling population dynamics in presence of hybridization between @SapienzaRoma \u0026amp; @umontpellier #WomenInSTEM #HMM #CaptureRecapture #MatrixModels #IndividualBasedModels #WildlifeManagement pic.twitter.com/grqiuP9xMG\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 21020 ","date":1582243200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1582243200,"objectID":"af872f2f3fec24d51ac936d8a8d8aca7","permalink":"https://oliviergimenez.github.io/blog/ninasdefense/","publishdate":"2020-02-21T00:00:00Z","relpermalink":"/blog/ninasdefense/","section":"blog","summary":"🇮🇹🇫🇷 @NSantostasi succesfully defended her PhD on modelling population dynamics in presence of hybridization between @SapienzaRoma \u0026amp; @umontpellier #WomenInSTEM #HMM #CaptureRecapture #MatrixModels #IndividualBasedModels #WildlifeManagement pic.twitter.com/grqiuP9xMG\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 21020 ","tags":["HMM","wolf demography","hybrids"],"title":"Nina's PhD defense on modelling population dynamics in presence of hybridization","type":"blog"},{"authors":null,"categories":null,"content":"#AcademicTwitter Thought I\u0026#39;d share this excellent thread in English explaining the devastating consequences of the reform of academia (universities and research institutes) and the pension reform in France - kuddos to @MHoussayH for this huge and worth effort 👏 https://t.co/q8TlnOIiKX\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 18020 ","date":1581984000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1581984000,"objectID":"91d1f326ea29724ac0d13723e20b6c31","permalink":"https://oliviergimenez.github.io/blog/reform/","publishdate":"2020-02-18T00:00:00Z","relpermalink":"/blog/reform/","section":"blog","summary":"#AcademicTwitter Thought I\u0026#39;d share this excellent thread in English explaining the devastating consequences of the reform of academia (universities and research institutes) and the pension reform in France - kuddos to @MHoussayH for this huge and worth effort 👏 https://t.","tags":["reform of academia"],"title":"Reform of academia","type":"blog"},{"authors":null,"categories":null,"content":"We had to dig out some #rstats #jags code to fit a structural equation capture-recapture model we developed w/ @SCubi25 in a 8-year old @ESAEcology paper (!), and it runs like a charm 🥳🍾#reproducibility ➡️ https://t.co/Te6BKnxgMF pic.twitter.com/78tWbOhqZD\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 6, 2020 ","date":1580947200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1580947200,"objectID":"2ef67a8e8de92e707d9a8b694855669b","permalink":"https://oliviergimenez.github.io/blog/semjags/","publishdate":"2020-02-06T00:00:00Z","relpermalink":"/blog/semjags/","section":"blog","summary":"We had to dig out some #rstats #jags code to fit a structural equation capture-recapture model we developed w/ @SCubi25 in a 8-year old @ESAEcology paper (!), and it runs like a charm 🥳🍾#reproducibility ➡️ https://t.","tags":["capture-recapture","SEM","rstats"],"title":"Bayesian SEM capture-recapture models","type":"blog"},{"authors":null,"categories":null,"content":"🇬🇷🐬📈@NSantostasi shows in her new paper https://t.co/EZIvJFZbV1 that Common dolphins in the Gulf of Corinth are Critically Endangered (w/ a stochastic PVA) #WomenInSTEM @IUCNRedList #FreeAccess 🇫🇷🇮🇹 pic.twitter.com/eC0plbVgeW\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 2020 ","date":1580601600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1580601600,"objectID":"9a4d6cd57ef01206c0c1de4774ad49b6","permalink":"https://oliviergimenez.github.io/blog/ninapva/","publishdate":"2020-02-02T00:00:00Z","relpermalink":"/blog/ninapva/","section":"blog","summary":"🇬🇷🐬📈@NSantostasi shows in her new paper https://t.co/EZIvJFZbV1 that Common dolphins in the Gulf of Corinth are Critically Endangered (w/ a stochastic PVA) #WomenInSTEM @IUCNRedList #FreeAccess 🇫🇷🇮🇹 pic.twitter.com/eC0plbVgeW\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 2020 ","tags":["PVA","Gulf of Corinth","rstats"],"title":"Dolphin PVA","type":"blog"},{"authors":null,"categories":null,"content":"Hello, everyone, this our #FirstTweet We are a group of @CNRS_OccitaniE and @EPHE_fr researchers working at the interface of #ecology, #modelling and #SocialSciences. HAIR is for Human-Animal InteRactions https://t.co/cCwVEmPaCE (new website to come soon) pic.twitter.com/zZX23MQ8L7\n\u0026mdash; twitt\u0026#39;hair (@twitthair1) January 28, 2020 ","date":1580169600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1580169600,"objectID":"6ddedb96b34f0ec37de1a82d0c31fd5d","permalink":"https://oliviergimenez.github.io/blog/team/","publishdate":"2020-01-28T00:00:00Z","relpermalink":"/blog/team/","section":"blog","summary":"Hello, everyone, this our #FirstTweet We are a group of @CNRS_OccitaniE and @EPHE_fr researchers working at the interface of #ecology, #modelling and #SocialSciences. HAIR is for Human-Animal InteRactions https://t.co/cCwVEmPaCE (new website to come soon) pic.","tags":["human-wildlife interactions","research team"],"title":"Our new team is on !","type":"blog"},{"authors":null,"categories":null,"content":"New @biorxivpreprint #preprint led by Sarah Bauduin 🐺💻 From ind behavior and pack dynamics to pop responses: An #IBM to model #wolf social life cycle https://t.co/ea31RvDMoE w/ contributions by @NSantostasi \u0026amp; @oksanagrente #rstats code ➡️ https://t.co/r0hSZYyTpe #WomenInSTEM pic.twitter.com/i8e0C9Ow8w\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) January 26, 2020 ","date":1579996800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1579996800,"objectID":"cf55a29e4a31bf3e30b322fecfd82d79","permalink":"https://oliviergimenez.github.io/blog/ibmwolf/","publishdate":"2020-01-26T00:00:00Z","relpermalink":"/blog/ibmwolf/","section":"blog","summary":"New @biorxivpreprint #preprint led by Sarah Bauduin 🐺💻 From ind behavior and pack dynamics to pop responses: An #IBM to model #wolf social life cycle https://t.co/ea31RvDMoE w/ contributions by @NSantostasi \u0026amp; @oksanagrente #rstats code ➡️ https://t.","tags":["IBM","wolf population dynamics","rstats"],"title":"Wolf population dynamics and individual-based models","type":"blog"},{"authors":null,"categories":null,"content":"Réunion animation du réseau loup/lynx pour l\u0026#39;#hérault #département34 @Occitanie à @villelodeve @OFBiodiversite - un effort de terrain conséquent https://t.co/d6kYRBuilD et précieux https://t.co/Vvem1it4S3 pour la recherche en écologie #SciencesCitoyennes pic.twitter.com/8TbvvotPp9\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) January 20, 2020 ","date":1579478400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1579478400,"objectID":"6a2a36be312a7b5c87ead8a34f254027","permalink":"https://oliviergimenez.github.io/blog/reseauloup/","publishdate":"2020-01-20T00:00:00Z","relpermalink":"/blog/reseauloup/","section":"blog","summary":"Réunion animation du réseau loup/lynx pour l\u0026#39;#hérault #département34 @Occitanie à @villelodeve @OFBiodiversite - un effort de terrain conséquent https://t.co/d6kYRBuilD et précieux https://t.co/Vvem1it4S3 pour la recherche en écologie #SciencesCitoyennes pic.twitter.com/8TbvvotPp9\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) January 20, 2020 ","tags":["loup","gestion des populations"],"title":"Réunion animation du réseau loup/lynx","type":"blog"},{"authors":null,"categories":null,"content":"We had the visit of Eivind F. Kleiven who\u0026#39;s working on small rodents in 🇳🇴 \u0026amp; the fascinating lemmings (awesome paper ➡️ https://t.co/9mPlFNbfd4) - another stimulating collaboration w/ N. Yoccoz, R. Ims \u0026amp; @FredBarraquand \u0026quot;we came to Norway for polar bears, we\u0026#39;ll stay for lemmings\u0026quot; pic.twitter.com/YxfWryID1S\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) January 17, 2020 ","date":1579219200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1579219200,"objectID":"547c2a7f0360600d3bd3f820344d48a8","permalink":"https://oliviergimenez.github.io/blog/eivindvisit/","publishdate":"2020-01-17T00:00:00Z","relpermalink":"/blog/eivindvisit/","section":"blog","summary":"We had the visit of Eivind F. Kleiven who\u0026#39;s working on small rodents in 🇳🇴 \u0026amp; the fascinating lemmings (awesome paper ➡️ https://t.co/9mPlFNbfd4) - another stimulating collaboration w/ N. Yoccoz, R.","tags":["lemmings","animal demography","lemmings","multispecies occupancy"],"title":"Eivind F. Kleiven's visit and lemmings pop dynamics","type":"blog"},{"authors":null,"categories":null,"content":"☘️🚜🍇 New #preprint led by @kazakou_elena in which we assess the effect of management practices on weed demography w/ hidden Markov models #HMM and #RandomEffects #WomenInSTEM #WomenInScience #OpenBUGS pic.twitter.com/xH5mQiMzd8\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) January 9, 2020 ","date":1578528000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1578528000,"objectID":"1d90c9ad6448184a0083f00a02086e33","permalink":"https://oliviergimenez.github.io/blog/elenispaper/","publishdate":"2020-01-09T00:00:00Z","relpermalink":"/blog/elenispaper/","section":"blog","summary":"☘️🚜🍇 New #preprint led by @kazakou_elena in which we assess the effect of management practices on weed demography w/ hidden Markov models #HMM and #RandomEffects #WomenInSTEM #WomenInScience #OpenBUGS pic.twitter.com/xH5mQiMzd8\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) January 9, 2020 ","tags":["HMM","plant demography","weed ecology"],"title":"Weed demography and hidden Markov models","type":"blog"},{"authors":null,"categories":null,"content":"Happy 2020 #rstats! To celebrate, my modest first #TidyTuesday submission for wk 2019-52. Spatio-temporal trends in 🐺 presence in 🇫🇷 w/ data from @oncfs @OFBiodiversite (bugged dynamic map below)#ggplot2 #tidyverse #dataviz #sf @thomas_mock Code: https://t.co/eKkehXlTJH pic.twitter.com/hHwkOmSamR\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) January 2, 2020 ","date":1577923200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1577923200,"objectID":"8f7532d5d369247862409697c65cdc8a","permalink":"https://oliviergimenez.github.io/blog/tidytuesday/","publishdate":"2020-01-02T00:00:00Z","relpermalink":"/blog/tidytuesday/","section":"blog","summary":"Happy 2020 #rstats! To celebrate, my modest first #TidyTuesday submission for wk 2019-52. Spatio-temporal trends in 🐺 presence in 🇫🇷 w/ data from @oncfs @OFBiodiversite (bugged dynamic map below)#ggplot2 #tidyverse #dataviz #sf @thomas_mock Code: https://t.","tags":["tidytuesday","rstats"],"title":"Tidytuesday","type":"blog"},{"authors":null,"categories":null,"content":"","date":1577836800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1577836800,"objectID":"8576ec274c98b3831668a172fa632d80","permalink":"https://oliviergimenez.github.io/about/","publishdate":"2020-01-01T00:00:00Z","relpermalink":"/about/","section":"","summary":"About","tags":null,"title":"About","type":"widget_page"},{"authors":null,"categories":null,"content":"","date":1577836800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1577836800,"objectID":"8e7bc052bdfc6746ea2bb6595e8093eb","permalink":"https://oliviergimenez.github.io/home/","publishdate":"2020-01-01T00:00:00Z","relpermalink":"/home/","section":"","summary":"About","tags":null,"title":"About","type":"widget_page"},{"authors":null,"categories":null,"content":"","date":1577836800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1577836800,"objectID":"3960dd3bdc6f629fb800d1d2aaa7224f","permalink":"https://oliviergimenez.github.io/resume/","publishdate":"2020-01-01T00:00:00Z","relpermalink":"/resume/","section":"","summary":"Resume","tags":null,"title":"Resume","type":"widget_page"},{"authors":null,"categories":null,"content":"Our team has had a fantastic 2019 year! Below is a thread w/ some highlights 🥳🍻📈 @LMarescot @NSantostasi @MaudQueroue @ValentinLauret @oksanagrente @GueryLorelei @roques_seb @SCubi25 @LouvrierJulie https://t.co/FaP9o0bOD4 pic.twitter.com/TuAVaAYkIT\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 29, 2019 2/7. We published papers on the management and #conservation of #wildlife wildlife resorting to #StatisticalEcology and human \u0026amp; social sciences https://t.co/Xlh6reUUQc @EcographyJourna @MethodsEcolEvol @Ecol_Evol @ConBiology @AnimalConserv @JAppliedEcology @biorxivpreprint pic.twitter.com/UbGFGJmmoQ\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 29, 2019 4/7. We engaged w/ stakeholders to try and have an impact on wildlife management. Several projects are on tracks in that direction https://t.co/R4UcwpDxVN pic.twitter.com/Y44sEcyhYb\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 29, 2019 6/7. We are on cloud 9 and lucky to have secured permanent positions in the team #SarahCubaynes (left) @EPHE_fr, #SarahBauduin (middle) @oncfs and #GillesMaurer (right) @zoobeauval pic.twitter.com/T7GRF9LFet\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 29, 2019 Happy New Year! pic.twitter.com/ZMMl4NX7ZK\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 29, 2019 ","date":1577577600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1577577600,"objectID":"9f92a15e23975e9c4688052159f79959","permalink":"https://oliviergimenez.github.io/blog/2019retrospective/","publishdate":"2019-12-29T00:00:00Z","relpermalink":"/blog/2019retrospective/","section":"blog","summary":"Our team has had a fantastic 2019 year! Below is a thread w/ some highlights 🥳🍻📈 @LMarescot @NSantostasi @MaudQueroue @ValentinLauret @oksanagrente @GueryLorelei @roques_seb @SCubi25 @LouvrierJulie https://t.co/FaP9o0bOD4 pic.twitter.com/TuAVaAYkIT\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 29, 2019 ","tags":["retrospective","boulet team"],"title":"Looking back at 2019","type":"blog"},{"authors":null,"categories":null,"content":"Procrastination... Je me suis \u0026quot;amusé\u0026quot; à reproduire cette géniale figure de @coulmont sur le jour des décès en 🇫🇷 en fonction de l\u0026#39;âge et du temps avec #rstats et le #tidyverse https://t.co/W8QNBX5WlT #ggplot2 #dplyr #lubridate https://t.co/wOthJOBzBd\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 23, 2019 ","date":1577059200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1577059200,"objectID":"7a3bb5d061fecf335036b1d7f13d77a5","permalink":"https://oliviergimenez.github.io/blog/procrastination_coulmont/","publishdate":"2019-12-23T00:00:00Z","relpermalink":"/blog/procrastination_coulmont/","section":"blog","summary":"Procrastination... Je me suis \u0026quot;amusé\u0026quot; à reproduire cette géniale figure de @coulmont sur le jour des décès en 🇫🇷 en fonction de l\u0026#39;âge et du temps avec #rstats et le #tidyverse https://t.","tags":["tidyverse","demographie"],"title":"Mortalité en France et tidyverse","type":"blog"},{"authors":null,"categories":null,"content":"We\u0026#39;ve been thinking a lot about #predictions in ecology lately. Here\u0026#39;s our modest crack at it https://t.co/NU7ROWLFWF led by @LouvrierJulie w/ a mechanistic-statistical #SpeciesDistributionModel to #forecast 🇫🇷🐺 expansion #ImperfectDetection #WomenInSTEM #ODEs #Bayes pic.twitter.com/D1n38edpDd\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 23, 2019 ","date":1577059200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1577059200,"objectID":"d4c711e4e0e3a35606c82f5584acb242","permalink":"https://oliviergimenez.github.io/blog/preprintmecastat/","publishdate":"2019-12-23T00:00:00Z","relpermalink":"/blog/preprintmecastat/","section":"blog","summary":"We\u0026#39;ve been thinking a lot about #predictions in ecology lately. Here\u0026#39;s our modest crack at it https://t.co/NU7ROWLFWF led by @LouvrierJulie w/ a mechanistic-statistical #SpeciesDistributionModel to #forecast 🇫🇷🐺 expansion #ImperfectDetection #WomenInSTEM #ODEs #Bayes pic.","tags":["SDM","wolf","management","prediction","forecasting"],"title":"New preprint by Julie Louvrier on forecasting species distribution","type":"blog"},{"authors":null,"categories":null,"content":"Playing around w/ awesome @R_nimble to fit multievent #HMM capture-recapture models to data - Code and data here https://t.co/2oHvsplahv Comments welcome 🤗 pic.twitter.com/tA89r80YYV\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 2019 ","date":1576800000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1576800000,"objectID":"1ee1fff67419804caa211258bd72966d","permalink":"https://oliviergimenez.github.io/blog/nimblehmm/","publishdate":"2019-12-20T00:00:00Z","relpermalink":"/blog/nimblehmm/","section":"blog","summary":"Playing around w/ awesome @R_nimble to fit multievent #HMM capture-recapture models to data - Code and data here https://t.co/2oHvsplahv Comments welcome 🤗 pic.twitter.com/tA89r80YYV\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 2019 ","tags":["HMM","Nimble","Bayes","MCMC","capture-recapture","multievent"],"title":"Nimble and HMM","type":"blog"},{"authors":null,"categories":null,"content":"Our team will be at the World Marine Mammal Conference in Barcelona @VisitBCN_FR 🐬📊🤩 @ValentinLauret @NSantostasi @GIS_3M #WMMC19 @wmmc2019 #MarineMammals - check out the who, what, when and where below pic.twitter.com/eaG36U4e9g\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 3, 2019 ","date":1575331200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1575331200,"objectID":"291bf36b9a93361eb43b6952f69a4f33","permalink":"https://oliviergimenez.github.io/blog/wmmc/","publishdate":"2019-12-03T00:00:00Z","relpermalink":"/blog/wmmc/","section":"blog","summary":"Our team will be at the World Marine Mammal Conference in Barcelona @VisitBCN_FR 🐬📊🤩 @ValentinLauret @NSantostasi @GIS_3M #WMMC19 @wmmc2019 #MarineMammals - check out the who, what, when and where below pic.","tags":["marine mammal","conservation biology"],"title":"World Marine Mammal Conference","type":"blog"},{"authors":null,"categories":null,"content":"This week we host our first winter school on «Reproducible Research in Numerical Ecology » co-organised by #GDR_Ecostat \u0026amp; #CESAB @FRBiodiv. Many thanks to our incredible list of speakers : Nicolas Casajus, Stéphane Dray, @GueryLorelei, @oaggimenez, @FGuilhaumon \u0026amp; @NinaSchiett !! pic.twitter.com/lRKlVXtYym\n\u0026mdash; Nicolas Mouquet (@NicolasMouquet) December 2,2019 ","date":1575244800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1575244800,"objectID":"e5674a0bbd3a91f9ae7e61346da11136","permalink":"https://oliviergimenez.github.io/blog/cesabworkshop/","publishdate":"2019-12-02T00:00:00Z","relpermalink":"/blog/cesabworkshop/","section":"blog","summary":"This week we host our first winter school on «Reproducible Research in Numerical Ecology » co-organised by #GDR_Ecostat \u0026amp; #CESAB @FRBiodiv. Many thanks to our incredible list of speakers : Nicolas Casajus, Stéphane Dray, @GueryLorelei, @oaggimenez, @FGuilhaumon \u0026amp; @NinaSchiett !","tags":["reproducible science","cesab","gdr ecostat","statistical ecology","tidyverse"],"title":"Workshop on reproducible science","type":"blog"},{"authors":null,"categories":null,"content":"New @biorxivpreprint #preprint in which @ValentinLauret combines aerial \u0026amp; boat surveys in #SDM for bottlenose dolphins w/ #occupancy models (single and repeated visits) - w/ H. Labach \u0026amp; M. Authier https://t.co/t2IJMTbUf8 pic.twitter.com/F2ftXu0DtK\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) November 21 2019 ","date":1574294400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1574294400,"objectID":"7004475982c7fd02ddc400458b945d20","permalink":"https://oliviergimenez.github.io/blog/preprintval/","publishdate":"2019-11-21T00:00:00Z","relpermalink":"/blog/preprintval/","section":"blog","summary":"New @biorxivpreprint #preprint in which @ValentinLauret combines aerial \u0026amp; boat surveys in #SDM for bottlenose dolphins w/ #occupancy models (single and repeated visits) - w/ H. Labach \u0026amp; M. Authier https://t.","tags":["SDM","tursiops","conservation","bottlenose dolphin","data integration"],"title":"New preprint by Valentin Lauret on integrated SDMs","type":"blog"},{"authors":null,"categories":null,"content":"Awesome Nimble @R_nimble workshop given by Daniel Turek from @WilliamsCollege #Bayes #StatisticalEcology at @CNRS_OccitaniE @INEE_CNRS @umontpellier w/ @oksanagrente @ValentinLauret @MaudQueroue @roques_seb @GSouchay @DavidVallecill_ @LoBacon1 pic.twitter.com/7zjQWKeF9P\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) October 23, 2019 ","date":1571788800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1571788800,"objectID":"fddad995a465f503e020e51d76a7869c","permalink":"https://oliviergimenez.github.io/blog/nimbleworkshop/","publishdate":"2019-10-23T00:00:00Z","relpermalink":"/blog/nimbleworkshop/","section":"blog","summary":"Awesome Nimble @R_nimble workshop given by Daniel Turek from @WilliamsCollege #Bayes #StatisticalEcology at @CNRS_OccitaniE @INEE_CNRS @umontpellier w/ @oksanagrente @ValentinLauret @MaudQueroue @roques_seb @GSouchay @DavidVallecill_ @LoBacon1 pic.twitter.com/7zjQWKeF9P\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) October 23, 2019 ","tags":["workshop","Nimble","Bayesian statistics"],"title":"Nimble workshop with Daniel Turek","type":"blog"},{"authors":null,"categories":null,"content":"Formation #CESAB / GDR EcoStat: Bonnes pratiques pour une recherche reproductible en écologie numérique, 2-6/12/2019, Montpellier. Par Nicolas Casajus, Stéphane Dray, @GueryLorelei, @oaggimenez @FGuilhaumon \u0026amp; @NinaSchiett\n➡https://t.co/P0HVLMEf7f@FRBiodiv @INEE_CNRS @Ifremer_fr pic.twitter.com/F9jCllLw6X\n\u0026mdash; Nicolas Mouquet (@NicolasMouquet) October 15, 2019 ","date":1571097600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1571097600,"objectID":"aac0b28f18a475f476f48aaf82ae4057","permalink":"https://oliviergimenez.github.io/blog/cesab/","publishdate":"2019-10-15T00:00:00Z","relpermalink":"/blog/cesab/","section":"blog","summary":"Formation #CESAB / GDR EcoStat: Bonnes pratiques pour une recherche reproductible en écologie numérique, 2-6/12/2019, Montpellier. Par Nicolas Casajus, Stéphane Dray, @GueryLorelei, @oaggimenez @FGuilhaumon \u0026amp; @NinaSchiett\n➡https://t.co/P0HVLMEf7f@FRBiodiv @INEE_CNRS @Ifremer_fr pic.twitter.com/F9jCllLw6X","tags":["workshop","ecostat","reproducible science"],"title":"Workshop on reproducible science","type":"blog"},{"authors":null,"categories":null,"content":"A thread on my review of the capture-recapture literature for the conference Wildlife Research and Conservation 2019 https://t.co/dqz8kbLdym in Berlin 🇩🇪. Slides, analyses, data \u0026amp; #rstats codes available from https://t.co/s9sFbkZG4k 1/23 pic.twitter.com/crGkVqYzov\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) September 28, /a\u003e ","date":1569628800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1569628800,"objectID":"903e1d7c8e43d1413172ab058242de98","permalink":"https://oliviergimenez.github.io/blog/crreview/","publishdate":"2019-09-28T00:00:00Z","relpermalink":"/blog/crreview/","section":"blog","summary":"A thread on my review of the capture-recapture literature for the conference Wildlife Research and Conservation 2019 https://t.co/dqz8kbLdym in Berlin 🇩🇪. Slides, analyses, data \u0026amp; #rstats codes available from https://t.co/s9sFbkZG4k 1/23 pic.","tags":["capture-recapture","systematic review"],"title":"A review on the last decade of research on capture-recapture","type":"blog"},{"authors":null,"categories":null,"content":"🥳🍾 Our paper providing lynx density estimates for 🇫🇷 Jura and Vosges mountains is out in @Ecol_Evol https://t.co/74rA1iUaQU @oncfs @CROC_Estelle @ZiFridolin #CameraTrap #SCR #spatial #capturerecapture w/ @oSCR_package #rstats #openaccess https://t.co/s8ooMzNDVI\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) September 28,2019 ","date":1569628800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1569628800,"objectID":"f7c073bb8a304365347ae652f9b595ee","permalink":"https://oliviergimenez.github.io/blog/paper_lynxdensities/","publishdate":"2019-09-28T00:00:00Z","relpermalink":"/blog/paper_lynxdensities/","section":"blog","summary":"🥳🍾 Our paper providing lynx density estimates for 🇫🇷 Jura and Vosges mountains is out in @Ecol_Evol https://t.co/74rA1iUaQU @oncfs @CROC_Estelle @ZiFridolin #CameraTrap #SCR #spatial #capturerecapture w/ @oSCR_package #rstats #openaccess https://t.co/s8ooMzNDVI","tags":["lynx","SCR models","population densities"],"title":"New paper on lynx densities in France","type":"blog"},{"authors":null,"categories":null,"content":"Découverte par les acteurs de l\u0026#39;interface logiciel pour évaluer l\u0026#39;impact d\u0026#39;aménagements sur la viabilité du lynx - 3ème atelier co-construction projet \u0026quot;Eviter, réduire \u0026amp; compenser la mortalité du Lynx par collision avec les véhicules de transport\u0026quot; https://t.co/uKY5ylOli2 #ITTECOP pic.twitter.com/0MWDBraOa3\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) September 25,2019 ","date":1569369600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1569369600,"objectID":"5c3f51e3014d2eff1a383a17d646b537","permalink":"https://oliviergimenez.github.io/blog/atelier3erclynx/","publishdate":"2019-09-25T00:00:00Z","relpermalink":"/blog/atelier3erclynx/","section":"blog","summary":"Découverte par les acteurs de l\u0026#39;interface logiciel pour évaluer l\u0026#39;impact d\u0026#39;aménagements sur la viabilité du lynx - 3ème atelier co-construction projet \u0026quot;Eviter, réduire \u0026amp; compenser la mortalité du Lynx par collision avec les véhicules de transport\u0026quot; https://t.","tags":["lynx","co-construction"],"title":"Co-construction d'une interface logiciel","type":"blog"},{"authors":null,"categories":null,"content":"New #preprint by @MGamelon on \u0026quot;Efficient use of harvest data: An integrated population model for exploited animal populations\u0026quot; https://t.co/s2nVkphPTe in which we combine data on dead and alive animals to infer population dynamics of the fascinating wild boar pic.twitter.com/MuJNXWjYZ5\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) September 20 2019 ","date":1568937600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1568937600,"objectID":"6d758f1ee39e00222a7be373444fef79","permalink":"https://oliviergimenez.github.io/blog/paper_ipmboar/","publishdate":"2019-09-20T00:00:00Z","relpermalink":"/blog/paper_ipmboar/","section":"blog","summary":"New #preprint by @MGamelon on \u0026quot;Efficient use of harvest data: An integrated population model for exploited animal populations\u0026quot; https://t.co/s2nVkphPTe in which we combine data on dead and alive animals to infer population dynamics of the fascinating wild boar pic.","tags":["IPM","wild boar"],"title":"New preprint: Efficient use of harvest data: An integrated population model for exploited animal populations","type":"blog"},{"authors":null,"categories":null,"content":"Very happy that our article on fitting species distribution models via combined likelihood to presence-only and occupancy data with lasso penalties and accounting for spatial dependence has been published in MEE! Go team! @LouvrierJulie @oaggimenez 🇦🇺🇫🇷🧀https://t.co/apVUbNb7Fi\n\u0026mdash; Ian Renner (@MoonbeamLevels) September 1819 ","date":1568764800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1568764800,"objectID":"8ee45ef39d4ccdd94e5c442a676e8782","permalink":"https://oliviergimenez.github.io/blog/sdmpaper/","publishdate":"2019-09-18T00:00:00Z","relpermalink":"/blog/sdmpaper/","section":"blog","summary":"Very happy that our article on fitting species distribution models via combined likelihood to presence-only and occupancy data with lasso penalties and accounting for spatial dependence has been published in MEE!","tags":["species distribution models","data combination"],"title":"New paper on SDMs and data combination","type":"blog"},{"authors":null,"categories":null,"content":"Stimulating 🇨🇭🇫🇷 meeting to discuss transboundary monitoring and research on 🐱🐺 in the Jura ⛰️ #KORA @oncfs (@INEE_CNRS) C. Duchamp, @ZiFridolin, U. Breitenmoser, C. Breitenmoser-Würsten, and N. Drouet-Hoguet not on the pix pic.twitter.com/WInB8uh1Vh\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) September 13, 20/a\u003e ","date":1568332800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1568332800,"objectID":"a309d6d210cd4dd04ea302a69eaa6de7","permalink":"https://oliviergimenez.github.io/blog/koraoncfsmeeting/","publishdate":"2019-09-13T00:00:00Z","relpermalink":"/blog/koraoncfsmeeting/","section":"blog","summary":"Stimulating 🇨🇭🇫🇷 meeting to discuss transboundary monitoring and research on 🐱🐺 in the Jura ⛰️ #KORA @oncfs (@INEE_CNRS) C. Duchamp, @ZiFridolin, U. Breitenmoser, C. Breitenmoser-Würsten, and N. Drouet-Hoguet not on the pix pic.","tags":["lynx","transboundary monitoring","large carnivores"],"title":"Lynx meeting on transboundary monitoring and research","type":"blog"},{"authors":null,"categories":null,"content":"🥳 It is a pleasure to welcome Daniel Turek @WilliamsCollege for a 6-month sabbatical here in @montpellier_ Stay tuned for more about #Bayes #StatisticalEcology w/ @R_nimble 🤩 Check out Daniel\u0026#39;s website ▶️ https://t.co/HYyMMrN6Nb pic.twitter.com/vK7q6bYzPU\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) September 919 ","date":1567987200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1567987200,"objectID":"491d8d4ee1463010d2b0f7529da82d80","permalink":"https://oliviergimenez.github.io/blog/daniel_turek/","publishdate":"2019-09-09T00:00:00Z","relpermalink":"/blog/daniel_turek/","section":"blog","summary":"🥳 It is a pleasure to welcome Daniel Turek @WilliamsCollege for a 6-month sabbatical here in @montpellier_ Stay tuned for more about #Bayes #StatisticalEcology w/ @R_nimble 🤩 Check out Daniel\u0026#39;s website ▶️ https://t.","tags":["sabbatical","Nimble","Bayes"],"title":"Daniel Turek is with us for 6 months!","type":"blog"},{"authors":null,"categories":null,"content":"New awesome paper by Sarah Bauduin https://t.co/Axu1kkkdSO in which she and her co-authors introduce #NetLogoR to build individual-based models w/ R #rstats 🥳🤩 Stay tuned as @oksanagrente and @NSantostasi are using #NetLogoR to study pop management and hybridization w/ wolf 🐺 https://t.co/fddUqENrAw\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) August 13, 2019 ","date":1565654400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1565654400,"objectID":"dc3de34c7b7571dc93a129bbe97be8c7","permalink":"https://oliviergimenez.github.io/blog/paper_netlogor/","publishdate":"2019-08-13T00:00:00Z","relpermalink":"/blog/paper_netlogor/","section":"blog","summary":"New awesome paper by Sarah Bauduin https://t.co/Axu1kkkdSO in which she and her co-authors introduce #NetLogoR to build individual-based models w/ R #rstats 🥳🤩 Stay tuned as @oksanagrente and @NSantostasi are using #NetLogoR to study pop management and hybridization w/ wolf 🐺 https://t.","tags":["R package","individual-based models"],"title":"Awesome paper introducing NetLogoR to build individual-based models with R","type":"blog"},{"authors":null,"categories":null,"content":"New #preprint w/ @GIS_3M on Distribution and abundance of bottlenose dolphin (Tursiops truncatus) over the French Mediterranean continental shelf https://t.co/eNfgLCT4k4 - Soooo impressed by the huge collective field effort to collect photo-id on this iconic species! pic.twitter.com/gRqpzkg6UK\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) August 3, 2019 ","date":1564790400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1564790400,"objectID":"fc50de92b800b14c7d41b25696afef72","permalink":"https://oliviergimenez.github.io/blog/paper_hlabach/","publishdate":"2019-08-03T00:00:00Z","relpermalink":"/blog/paper_hlabach/","section":"blog","summary":"New #preprint w/ @GIS_3M on Distribution and abundance of bottlenose dolphin (Tursiops truncatus) over the French Mediterranean continental shelf https://t.co/eNfgLCT4k4 - Soooo impressed by the huge collective field effort to collect photo-id on this iconic species!","tags":["dolphins","tursiops","capture-recapture","abundance"],"title":"New preprint: Distribution and abundance of bottlenose dolphin (Tursiops truncatus) over the French Mediterranean continental shelf","type":"blog"},{"authors":null,"categories":null,"content":"Le projet porté par Gilles Maurer dans lequel nous étudions les conditions de la #coexistence entre humains et #éléphant d\u0026#39;Asie mis à l\u0026#39;honneur par @IsiteMUSE @umontpellier - collaborations entre @zoobeauval @INEE_CNRS @CNRS_OccitaniE @INSHS_CNRS pic.twitter.com/VvAYsYWDSc\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 19 2019 ","date":1563494400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1563494400,"objectID":"09bac254ab408729833065137b8e1612","permalink":"https://oliviergimenez.github.io/blog/peace/","publishdate":"2019-07-19T00:00:00Z","relpermalink":"/blog/peace/","section":"blog","summary":"Le projet porté par Gilles Maurer dans lequel nous étudions les conditions de la #coexistence entre humains et #éléphant d\u0026#39;Asie mis à l\u0026#39;honneur par @IsiteMUSE @umontpellier - collaborations entre @zoobeauval @INEE_CNRS @CNRS_OccitaniE @INSHS_CNRS pic.","tags":["éléphant","Muse","PEACE"],"title":"Gilles et le projet Muse sur la conservation de l'éléphant d'Asie","type":"blog"},{"authors":null,"categories":null,"content":"Excellent article \u0026quot;Doit-on (vraiment) crier au loup\u0026quot; par @CJosset @LEXPRESS - @oncfs @INEE_CNRS @CNRS_OccitaniE @IsiteMUSE @umontpellier https://t.co/PpmvNnpM9Y\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 15, 2019 ","date":1563148800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1563148800,"objectID":"d700c33a08d916d29050ac54e84ea235","permalink":"https://oliviergimenez.github.io/blog/article_loup_lexpress/","publishdate":"2019-07-15T00:00:00Z","relpermalink":"/blog/article_loup_lexpress/","section":"blog","summary":"Excellent article \u0026quot;Doit-on (vraiment) crier au loup\u0026quot; par @CJosset @LEXPRESS - @oncfs @INEE_CNRS @CNRS_OccitaniE @IsiteMUSE @umontpellier https://t.co/PpmvNnpM9Y\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 15, 2019 ","tags":["loup","presse"],"title":"Article sur le loup dans L'Express","type":"blog"},{"authors":null,"categories":null,"content":"🥳🍾 Awesome work by AS Bonnet Lebrun, A Karamanlidis from @Arcturosgr M de Gabriel Hernando and I Renner @MoonbeamLevels in which we use Poisson point process models to identify priority conservation areas for brown bears in Greece 🇬🇷🐻📈 https://t.co/rDV7h9Zb7C pic.twitter.com/UmgmynE3Jd\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 5 juillet19 ","date":1562284800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1562284800,"objectID":"c6744c1dd6fb6b98f996e26e7c119715","permalink":"https://oliviergimenez.github.io/blog/paperanneso/","publishdate":"2019-07-05T00:00:00Z","relpermalink":"/blog/paperanneso/","section":"blog","summary":"🥳🍾 Awesome work by AS Bonnet Lebrun, A Karamanlidis from @Arcturosgr M de Gabriel Hernando and I Renner @MoonbeamLevels in which we use Poisson point process models to identify priority conservation areas for brown bears in Greece 🇬🇷🐻📈 https://t.","tags":["point process","species distribution model","brown bear","large carnivores","Greece"],"title":"New paper: Identifying priority conservation areas for a recovering brown bear population in Greece using citizen science data","type":"blog"},{"authors":null,"categories":null,"content":"New paper by @TamarLok 🤩👏 using integrated modelling of 8 years of resighting and count data and accouting for heterogeneity to show the key importance of the Luannan Coast of Bohai Bay (China) as a spring #stopover area for red knots https://t.co/EgOA6yM4Dp #WomenInSTEM pic.twitter.com/05qVXhyXVt\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 28 juin 2019 ","date":1561680000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1561680000,"objectID":"4989c5f0dbaed9dee071d320cea1f317","permalink":"https://oliviergimenez.github.io/blog/papertamar/","publishdate":"2019-06-28T00:00:00Z","relpermalink":"/blog/papertamar/","section":"blog","summary":"New paper by @TamarLok 🤩👏 using integrated modelling of 8 years of resighting and count data and accouting for heterogeneity to show the key importance of the Luannan Coast of Bohai Bay (China) as a spring #stopover area for red knots https://t.","tags":["heterogeneity","demography","capture-recapture"],"title":"New paper: Accounting for heterogeneity when estimating stopover duration, timing and population size of red knots along the Luannan Coast of Bohai Bay, China","type":"blog"},{"authors":null,"categories":null,"content":"Awesome talk by @hgcaceres on \u0026#39;Bottom-up conservation, engaging with the local community to develop better conservation actions\u0026#39; 🤩 @CNRS_OccitaniE @INEE_CNRS @umontpellier @IsiteMUSE pic.twitter.com/dGOGFY9B8x\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 18juin 2019 ","date":1560643200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1560643200,"objectID":"085b23e2409486fd8b0fc2e1ba21a371","permalink":"https://oliviergimenez.github.io/blog/hernanvisit/","publishdate":"2019-06-16T00:00:00Z","relpermalink":"/blog/hernanvisit/","section":"blog","summary":"Awesome talk by @hgcaceres on \u0026#39;Bottom-up conservation, engaging with the local community to develop better conservation actions\u0026#39; 🤩 @CNRS_OccitaniE @INEE_CNRS @umontpellier @IsiteMUSE pic.twitter.com/dGOGFY9B8x\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 18juin 2019 ","tags":["adaptive management","network"],"title":"Seminar by Hernan Caceres on Bottom-up conservation, engaging with the local community to develop better conservation actions","type":"blog"},{"authors":null,"categories":null,"content":"You have gaps in your data time-series? Open-pop SCR models may help you infer abundance, even during the gap years @PDupont17 @joechip90 @oaggimenez @R_nimble 🚀@bxv_ecol @biorxiv_ecology https://t.co/0YjWa1PmIL pic.twitter.com/Aa4cQPAcoP\n\u0026mdash; Cyril Milleret (@Milleret_Cyril) 15 juin 2019 ","date":1560556800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1560556800,"objectID":"7f59b693b0b66c9f0c8f6216b7cce29b","permalink":"https://oliviergimenez.github.io/blog/preprintcyril/","publishdate":"2019-06-15T00:00:00Z","relpermalink":"/blog/preprintcyril/","section":"blog","summary":"You have gaps in your data time-series? Open-pop SCR models may help you infer abundance, even during the gap years @PDupont17 @joechip90 @oaggimenez @R_nimble 🚀@bxv_ecol @biorxiv_ecology https://t.co/0YjWa1PmIL pic.twitter.com/Aa4cQPAcoP\n\u0026mdash; Cyril Milleret (@Milleret_Cyril) 15 juin 2019 ","tags":["scr","large carnivores","spatial capture-recapture"],"title":"New preprint: Estimating abundance with interruptions in data collection using open population spatial capture-recapture models","type":"blog"},{"authors":null,"categories":null,"content":"Wow, the 5th conference of our 🇫🇷 group in #StatisticalEcology is open in the beautiful city of #Avignon @UnivAvignon #GdREcoStat - Check out the program ▶️ https://t.co/1H8z4VRGxI pic.twitter.com/2Fyw1NleJG\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 13 mai 2019 ","date":1557705600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1557705600,"objectID":"f723b2d4f8145203f2c5a6841ec60589","permalink":"https://oliviergimenez.github.io/blog/gdrecostat2019/","publishdate":"2019-05-13T00:00:00Z","relpermalink":"/blog/gdrecostat2019/","section":"blog","summary":"Wow, the 5th conference of our 🇫🇷 group in #StatisticalEcology is open in the beautiful city of #Avignon @UnivAvignon #GdREcoStat - Check out the program ▶️ https://t.co/1H8z4VRGxI pic.twitter.com/2Fyw1NleJG\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 13 mai 2019 ","tags":["statistical ecology","Avignon","GDR"],"title":"5th French conference on Statistical Ecology (thread)","type":"blog"},{"authors":null,"categories":null,"content":"I thought I\u0026#39;d share some #rstats code about local minima in #multistate #capturerecapture models https://t.co/aJE33dBtdO using #Mark #Admb #Jags #Stan (#Nimble \u0026amp; #TMB to come). Largely inspired by Chapter 9 of the \u0026quot;Gentle introduction to Mark\u0026quot; by Cooch \u0026amp; White 😉 pic.twitter.com/NxzG6w0Bzq\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 12 mai 2019 ","date":1557619200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1557619200,"objectID":"e43d5bcac59a60132d8f2347c3135857","permalink":"https://oliviergimenez.github.io/blog/localminima_reboot/","publishdate":"2019-05-12T00:00:00Z","relpermalink":"/blog/localminima_reboot/","section":"blog","summary":"I thought I\u0026#39;d share some #rstats code about local minima in #multistate #capturerecapture models https://t.co/aJE33dBtdO using #Mark #Admb #Jags #Stan (#Nimble \u0026amp; #TMB to come). Largely inspired by Chapter 9 of the \u0026quot;Gentle introduction to Mark\u0026quot; by Cooch \u0026amp; White 😉 pic.","tags":["local minima","capture-recapture"],"title":"Local minima and multistate capture-recapture models","type":"blog"},{"authors":null,"categories":null,"content":"Just wrote a short #tutorial on fitting a system of Ordinary Differential Equations to time-series data using #OpenBUGS in #rstats https://t.co/7luj8XVgP6 - Data and model are from https://t.co/X9CwX7dW83 in which #zombie movies are used for illustration #shaunofthedead pic.twitter.com/MHmiGhH7Ri\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 9 mai 2019 ","date":1557446400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1557446400,"objectID":"f74a377e029288907cce795d8126f188","permalink":"https://oliviergimenez.github.io/blog/fitodedata/","publishdate":"2019-05-10T00:00:00Z","relpermalink":"/blog/fitodedata/","section":"blog","summary":"Just wrote a short #tutorial on fitting a system of Ordinary Differential Equations to time-series data using #OpenBUGS in #rstats https://t.co/7luj8XVgP6 - Data and model are from https://t.co/X9CwX7dW83 in which #zombie movies are used for illustration #shaunofthedead pic.","tags":["ordinary differential equations","OpenBUGS","zombies"],"title":"Tutorial on fitting a system of Ordinary Differential Equations to time-series data using OpenBUGS","type":"blog"},{"authors":null,"categories":null,"content":"A nice post by @ksuryawanshi on our recent paper on transboundary management of wolverines in Scandinavia: \u0026quot;Carnivores without borders: management of transboundary populations when objectives differ\u0026quot; https://t.co/OcAAgwUP4U via @JAppliedEcology\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 9 mai 2019 ","date":1557360000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1557360000,"objectID":"0517a35d02b7663ba51a0fcc42c3fd99","permalink":"https://oliviergimenez.github.io/blog/blogpost_articlevincenzo/","publishdate":"2019-05-09T00:00:00Z","relpermalink":"/blog/blogpost_articlevincenzo/","section":"blog","summary":"A nice post by @ksuryawanshi on our recent paper on transboundary management of wolverines in Scandinavia: \u0026quot;Carnivores without borders: management of transboundary populations when objectives differ\u0026quot; https://t.co/OcAAgwUP4U via @JAppliedEcology\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 9 mai 2019 ","tags":["wolverine","blog post","transboundary management","large carnivores"],"title":"Carnivores without borders: management of transboundary populations when objectives differ","type":"blog"},{"authors":null,"categories":null,"content":"New paper on \u0026#39;How many cubs can a mum nurse? Maternal age and size influence litter size in polar bears\u0026#39; https://t.co/Utu4VZ0Ms3 w/ #DorindaFolio (former Master student 👏🤩) and #SarahCubaynes in control #WomenInSTEM - PDF on request 😉 pic.twitter.com/QABpoQxmHQ\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 1 mai 2019 ","date":1556668800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1556668800,"objectID":"45ec11e2bcd5b271836052a92b57f8c9","permalink":"https://oliviergimenez.github.io/blog/paperdorinda/","publishdate":"2019-05-01T00:00:00Z","relpermalink":"/blog/paperdorinda/","section":"blog","summary":"New paper on \u0026#39;How many cubs can a mum nurse? Maternal age and size influence litter size in polar bears\u0026#39; https://t.co/Utu4VZ0Ms3 w/ #DorindaFolio (former Master student 👏🤩) and #SarahCubaynes in control #WomenInSTEM - PDF on request 😉 pic.","tags":["polar bear","demography","life-history strategies"],"title":"New paper: How many cubs can a mum nurse? Maternal age and size influence litter size in polar bears","type":"blog"},{"authors":null,"categories":null,"content":"New #preprint on \u0026quot;Combining multiple data sources in species distribution models while accounting for spatial dependence and overfitting with penalised likelihood maximisation\u0026quot; https://t.co/7GLRrgpo4D w/ @MoonbeamLevels \u0026amp; @LouvrierJulie #SDM #MachineLearning #PointProcess pic.twitter.com/6xzAlARV9E\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 24 avri 2019 ","date":1556064000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1556064000,"objectID":"abe5152e886e8e0718c000280bcdb663","permalink":"https://oliviergimenez.github.io/blog/preprintian/","publishdate":"2019-04-24T00:00:00Z","relpermalink":"/blog/preprintian/","section":"blog","summary":"New #preprint on \u0026quot;Combining multiple data sources in species distribution models while accounting for spatial dependence and overfitting with penalised likelihood maximisation\u0026quot; https://t.co/7GLRrgpo4D w/ @MoonbeamLevels \u0026amp; @LouvrierJulie #SDM #MachineLearning #PointProcess pic.","tags":["sdm","occupancy","point process","lasso"],"title":"Combining multiple data sources in species distribution models while accounting for spatial dependence and overfitting with penalised likelihood maximisation","type":"blog"},{"authors":null,"categories":null,"content":"Quick and dirty code to fit a multistate single-season #occupancy model in Jags using a #HMM formulation, in case it\u0026#39;s useful https://t.co/vUlrP0jjIU #JAGS #rstats w/ @ArnaudLyet pic.twitter.com/dSr0BlICb6\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 13l 2019 ","date":1555113600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1555113600,"objectID":"c9a2ec26901b631444573711d522d9e0","permalink":"https://oliviergimenez.github.io/blog/coding/","publishdate":"2019-04-13T00:00:00Z","relpermalink":"/blog/coding/","section":"blog","summary":"Quick and dirty code to fit a multistate single-season #occupancy model in Jags using a #HMM formulation, in case it\u0026#39;s useful https://t.co/vUlrP0jjIU #JAGS #rstats w/ @ArnaudLyet pic.twitter.com/dSr0BlICb6\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 13l 2019 ","tags":["occupancy","jags"],"title":"Jags code to fit a multistate single-season occupancy model","type":"blog"},{"authors":null,"categories":null,"content":"New paper on \u0026quot;Inferring animal social networks with imperfect detection\u0026quot; w/ colleagues from 🇦🇷 and #LorenaMansilla our 🇨🇱 PhD student #SocialNetworks #CaptureRecapture #StateSpaceModel #rstats #BayesianStatistics - Download PDF freely here https://t.co/heixxmRdFH pic.twitter.com/dZbRLvFbjT\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 10 avril 2019 ","date":1554854400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1554854400,"objectID":"2e7ed5cb8c90c7fa549a0a83dbdfa00d","permalink":"https://oliviergimenez.github.io/blog/paperog/","publishdate":"2019-04-10T00:00:00Z","relpermalink":"/blog/paperog/","section":"blog","summary":"New paper on \u0026quot;Inferring animal social networks with imperfect detection\u0026quot; w/ colleagues from 🇦🇷 and #LorenaMansilla our 🇨🇱 PhD student #SocialNetworks #CaptureRecapture #StateSpaceModel #rstats #BayesianStatistics - Download PDF freely here https://t.","tags":["social network","dolphin","hmm"],"title":"New paper on Inferring animal social networks with imperfect detection","type":"blog"},{"authors":null,"categories":null,"content":"🐺🔎💩 Un chouette article https://t.co/Qm0fkSX675 dans @maglarecherche sur les travaux de thèse de @LouvrierJulie sur écologie statistique, modèles de distribution d\u0026#39;espèce et grands carnivores - merci @BereniceRbt ! Bel exemple de collaborations @oncfs @INEE_CNRS @umontpellier pic.twitter.com/xkvjC4hbQr\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 7l 2019 ","date":1554595200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1554595200,"objectID":"78c39a02602cdecc901fc48e3c2d9a8f","permalink":"https://oliviergimenez.github.io/blog/commloup/","publishdate":"2019-04-07T00:00:00Z","relpermalink":"/blog/commloup/","section":"blog","summary":"🐺🔎💩 Un chouette article https://t.co/Qm0fkSX675 dans @maglarecherche sur les travaux de thèse de @LouvrierJulie sur écologie statistique, modèles de distribution d\u0026#39;espèce et grands carnivores - merci @BereniceRbt ! Bel exemple de collaborations @oncfs @INEE_CNRS @umontpellier pic.","tags":["loup","occupancy"],"title":"Un article dans La Recherche sur nos travaux","type":"blog"},{"authors":null,"categories":null,"content":"How many lynx are there? \u0026quot;Spatial density estimates of Eurasian lynx (Lynx lynx) in the 🇫🇷 Jura and Vosges 🏔️\u0026quot; https://t.co/roLlOsFxw7 @oncfs @CROC_Estelle @ZiFridolin w/ awesome #oSCR @rstats 📦 @chrissuthy @andyroyle_pwrc @dwlinden @CNRS_OccitaniE @INEE_CNRS pic.twitter.com/AsUGJVhmKx\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 5 avril 2019 ","date":1554422400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1554422400,"objectID":"872f7082c34bdd54ccaf34412c5c5e26","permalink":"https://oliviergimenez.github.io/blog/preprintog/","publishdate":"2019-04-05T00:00:00Z","relpermalink":"/blog/preprintog/","section":"blog","summary":"How many lynx are there? \u0026quot;Spatial density estimates of Eurasian lynx (Lynx lynx) in the 🇫🇷 Jura and Vosges 🏔️\u0026quot; https://t.co/roLlOsFxw7 @oncfs @CROC_Estelle @ZiFridolin w/ awesome #oSCR @rstats 📦 @chrissuthy @andyroyle_pwrc @dwlinden @CNRS_OccitaniE @INEE_CNRS pic.","tags":["SCR","lynx"],"title":"Preprint on how many lynx in France","type":"blog"},{"authors":null,"categories":null,"content":"@FredBarraquand is on 🔥 - check out his new #arXiv preprint \u0026quot;Fitting stochastic predator-prey models using both population density and kill rate data\u0026quot; https://t.co/u23GoSoo81 pic.twitter.com/0YwHWbNepb\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 4 avril 2019 ","date":1554336000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1554336000,"objectID":"e37da58d188f01e24aa480a179e7d3b1","permalink":"https://oliviergimenez.github.io/blog/preprintfred2/","publishdate":"2019-04-04T00:00:00Z","relpermalink":"/blog/preprintfred2/","section":"blog","summary":"@FredBarraquand is on 🔥 - check out his new #arXiv preprint \u0026quot;Fitting stochastic predator-prey models using both population density and kill rate data\u0026quot; https://t.co/u23GoSoo81 pic.twitter.com/0YwHWbNepb\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 4 avril 2019 ","tags":["Democom","multispeces"],"title":"Fitting stochastic predator-prey models using both population density and kill rate data","type":"blog"},{"authors":null,"categories":null,"content":"🥳🧐 New preprint w/ @FredBarraquand \u0026quot;Integrating multiple data sources to fit matrix population models for interacting species\u0026quot; https://t.co/LSSqOcj0Ca w/ codes https://t.co/Msn8XnChu6; check out our #democom project for more https://t.co/eNFOzrPgWR #rstats #IPMs #predatorprey pic.twitter.com/kJdX16rbU6\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 2 avril 2019 ","date":1554163200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1554163200,"objectID":"a40b07f73bdab26d523de99606dc07e2","permalink":"https://oliviergimenez.github.io/blog/preprintfred1/","publishdate":"2019-04-02T00:00:00Z","relpermalink":"/blog/preprintfred1/","section":"blog","summary":"🥳🧐 New preprint w/ @FredBarraquand \u0026quot;Integrating multiple data sources to fit matrix population models for interacting species\u0026quot; https://t.co/LSSqOcj0Ca w/ codes https://t.co/Msn8XnChu6; check out our #democom project for more https://t.co/eNFOzrPgWR #rstats #IPMs #predatorprey pic.","tags":["Democom","ipm","multispeces"],"title":"Integrating multiple data sources to fit matrix population models for interacting species","type":"blog"},{"authors":null,"categories":null,"content":"New preprint by awesome #SarahCubaynes \u0026quot;Modeling the demography of species providing extended parental care: A #capturerecapture approach with a case study on #Polar Bears\u0026quot; https://t.co/x1H0X7GyKn #demography #rstats #multistate #WomenInSTEM #WomenInScience pic.twitter.com/rYEwOUobEn\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 2 avril 2019 ","date":1554163200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1554163200,"objectID":"4887cc190894d3c67d34ab8136da928d","permalink":"https://oliviergimenez.github.io/blog/preprintsarah/","publishdate":"2019-04-02T00:00:00Z","relpermalink":"/blog/preprintsarah/","section":"blog","summary":"New preprint by awesome #SarahCubaynes \u0026quot;Modeling the demography of species providing extended parental care: A #capturerecapture approach with a case study on #Polar Bears\u0026quot; https://t.co/x1H0X7GyKn #demography #rstats #multistate #WomenInSTEM #WomenInScience pic.","tags":["polar bear","demography","hmm"],"title":"Modeling the demography of species providing extended parental care: A capturerecapture approach with a case study on Polar Bears","type":"blog"},{"authors":null,"categories":null,"content":"New paper by #VincenzoGervasi \u0026quot;Failure to coordinate management in transboundary populations hinders the achievement of national management goals: The case of wolverines in Scandinavia\u0026quot; w/ @johnlinnell #HenrikBroseth #LabexCemeb @CNRS_OccitaniE @NINAforskning @umontpellier pic.twitter.com/gzkANotqry\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 2 avril 2019 ","date":1554163200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1554163200,"objectID":"05498fd80749f677327b726d557dec87","permalink":"https://oliviergimenez.github.io/blog/papervincenzo/","publishdate":"2019-04-02T00:00:00Z","relpermalink":"/blog/papervincenzo/","section":"blog","summary":"New paper by #VincenzoGervasi \u0026quot;Failure to coordinate management in transboundary populations hinders the achievement of national management goals: The case of wolverines in Scandinavia\u0026quot; w/ @johnlinnell #HenrikBroseth #LabexCemeb @CNRS_OccitaniE @NINAforskning @umontpellier pic.","tags":["wolverine","demography"],"title":"New paper: Failure to coordinate management in transboundary populations hinders the achievement of national management goals: The case of wolverines in Scandinavia","type":"blog"},{"authors":null,"categories":null,"content":"🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩 @CNRS_OccitaniE @INEE_CNRS @CNRS @CollegeDoctoUM @IsiteMUSE @umontpellier https://t.co/6JlClESHo5\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 13 février 2019 ","date":1550016000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1550016000,"objectID":"d4160f7aeba9fd8a0da334c1e88cd554","permalink":"https://oliviergimenez.github.io/blog/prixlarecherche/","publishdate":"2019-02-13T00:00:00Z","relpermalink":"/blog/prixlarecherche/","section":"blog","summary":"🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩 @CNRS_OccitaniE @INEE_CNRS @CNRS @CollegeDoctoUM @IsiteMUSE @umontpellier https://t.co/6JlClESHo5\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 13 février 2019 ","tags":["La Recherche","wolf"],"title":"Julie obtient le prix du magazine La Recherche","type":"blog"},{"authors":null,"categories":null,"content":"👩 💻🗺️👨 💻 The slides of my introduction to #GIS and #mapping in #rstats using the #sf 📦 and brown 🐻 distribution in the #pyrenees as a case study https://t.co/SKQOCzbxHn - raw material on #github https://t.co/dHoMz6I2Kp #rspatial #spatial #ggplot2 #tidyverse #dataviz pic.twitter.com/22eD1Y55d3\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 8 février 2019 ","date":1549584000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1549584000,"objectID":"121c6a2612fbf0daa64360277c0106e1","permalink":"https://oliviergimenez.github.io/blog/intro_spatial/","publishdate":"2019-02-08T00:00:00Z","relpermalink":"/blog/intro_spatial/","section":"blog","summary":"👩 💻🗺️👨 💻 The slides of my introduction to #GIS and #mapping in #rstats using the #sf 📦 and brown 🐻 distribution in the #pyrenees as a case study https://t.co/SKQOCzbxHn - raw material on #github https://t.","tags":["GIS","R","sf","ggplot","tidyverse"],"title":"Introduction to spatial analyses in R","type":"blog"},{"authors":null,"categories":null,"content":"Meeting of our Interact project https://t.co/Le2mkruYag on the interactions between bottlenose 🐬 and human activities in the French Mediterranean 🌊 @Fondationfrance @GIS_3M #EcoOcéanInstitut #SouffleursdEcume #Gecem @LIENSsCNRSULR @WWFFrance @WWFFrance_Press @CNRS_OccitaniE pic.twitter.com/hAJPnumxAr\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 6 février 2019 ","date":1549411200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1549411200,"objectID":"ec6b39e4ce3f38fa64070cef934e6c0a","permalink":"https://oliviergimenez.github.io/blog/interact_meeting2/","publishdate":"2019-02-06T00:00:00Z","relpermalink":"/blog/interact_meeting2/","section":"blog","summary":"Meeting of our Interact project https://t.co/Le2mkruYag on the interactions between bottlenose 🐬 and human activities in the French Mediterranean 🌊 @Fondationfrance @GIS_3M #EcoOcéanInstitut #SouffleursdEcume #Gecem @LIENSsCNRSULR @WWFFrance @WWFFrance_Press @CNRS_OccitaniE pic.twitter.com/hAJPnumxAr","tags":["meeting","interact"],"title":"Meeting on interactions between dolphins and human activities","type":"blog"},{"authors":null,"categories":null,"content":"Procrastination at its highest level 😜 My first attempt to design hex stickers for our #R2ucare 📦 https://t.co/ZbclwJKB5W #rstats code on #github https://t.co/TbEgjQf7iC Comments more than welcome 😁 pic.twitter.com/vVdVnJ9B79\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 6 février 2019 ","date":1549411200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1549411200,"objectID":"dab433d44c3fe2741e57bb62f7a3cb44","permalink":"https://oliviergimenez.github.io/blog/procrastination/","publishdate":"2019-02-06T00:00:00Z","relpermalink":"/blog/procrastination/","section":"blog","summary":"Procrastination at its highest level 😜 My first attempt to design hex stickers for our #R2ucare 📦 https://t.co/ZbclwJKB5W #rstats code on #github https://t.co/TbEgjQf7iC Comments more than welcome 😁 pic.twitter.com/vVdVnJ9B79\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 6 février 2019 ","tags":["logo","R","R2ucare"],"title":"Procrastination...","type":"blog"},{"authors":null,"categories":null,"content":"How many 🐺🐶 hybrids are there?! @NSantostasi shows how to estimate prevalence w/ #HMM imperfect detection in her @Ecol_Evol #Euring17 paper https://t.co/KwiHB2ZBpX #openaccess 🤝 work w/ @SapienzaRoma @ISPRA_Press @wolfappcenter @umontpellier @CNRS_OccitaniE @INEE_CNRS 🇮🇹🇫🇷 pic.twitter.com/jQJtBwNOZn\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 5 février 2019 ","date":1549324800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1549324800,"objectID":"94bd070e161a2de6acee5fc0709352cf","permalink":"https://oliviergimenez.github.io/blog/paper_nina/","publishdate":"2019-02-05T00:00:00Z","relpermalink":"/blog/paper_nina/","section":"blog","summary":"How many 🐺🐶 hybrids are there?! @NSantostasi shows how to estimate prevalence w/ #HMM imperfect detection in her @Ecol_Evol #Euring17 paper https://t.co/KwiHB2ZBpX #openaccess 🤝 work w/ @SapienzaRoma @ISPRA_Press @wolfappcenter @umontpellier @CNRS_OccitaniE @INEE_CNRS 🇮🇹🇫🇷 pic.","tags":["hybrid prevalence","democom","hmm"],"title":"How to estimate the prevalence of hybrids","type":"blog"},{"authors":null,"categories":null,"content":"Check out our new #bioRxiv preprint on the influence of maternal traits on litter size in #polar #bears led by former Master student #DorindaFolio \u0026quot;How many cubs can a mum raise? Maternal age and size influence litter size in polar bears\u0026quot; https://t.co/kJymSuIkMZ #WomenInSTEM pic.twitter.com/rrCYScxmlj\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 29 janvier 2019 ","date":1548720000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1548720000,"objectID":"7b31e5f2dc952252a31beb03c23dd9a8","permalink":"https://oliviergimenez.github.io/blog/preprint_dorinda/","publishdate":"2019-01-29T00:00:00Z","relpermalink":"/blog/preprint_dorinda/","section":"blog","summary":"Check out our new #bioRxiv preprint on the influence of maternal traits on litter size in #polar #bears led by former Master student #DorindaFolio \u0026quot;How many cubs can a mum raise?","tags":["life-history","polar bear"],"title":"How many cubs can a mum raise? Maternal age and size influence litter size in polar bears","type":"blog"},{"authors":null,"categories":null,"content":"My introduction to the #tidyverse for our lab meeting to manipulate and visualise data in #rstat https://t.co/As9bkXY9GZ. Feel free to steal and modify this material for your own use. Be advised, this is work in progress \u0026amp; a mix of 🇬🇧/🇫🇷😋 Comments welcome! #datascience #davaviz pic.twitter.com/2vrrdbzuWh\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 16 janvier 2019 ","date":1547596800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1547596800,"objectID":"a42109475dbc03270996c768573f491e","permalink":"https://oliviergimenez.github.io/blog/intro_tidyverse/","publishdate":"2019-01-16T00:00:00Z","relpermalink":"/blog/intro_tidyverse/","section":"blog","summary":"My introduction to the #tidyverse for our lab meeting to manipulate and visualise data in #rstat https://t.co/As9bkXY9GZ. Feel free to steal and modify this material for your own use. Be advised, this is work in progress \u0026amp; a mix of 🇬🇧/🇫🇷😋 Comments welcome!","tags":["tidyverse","R","data wrangling"],"title":"Introduction to the Tidyverse","type":"blog"},{"authors":null,"categories":null,"content":"Check out our new @biorxivpreprint #preprint \u0026quot;Inferring #wildlife #poaching in Southeast Asia with multispecies dynamic #occupancy models\u0026quot; led by @LMarescot 🐯 https://t.co/g4NAQnUODw #HMMs @ArnaudLyet @rhohits #NeilCarter @SMARTCnsvTools @AgenceRecherche pic.twitter.com/5EttLyY1S9\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 15 janvier 2019 ","date":1547510400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1547510400,"objectID":"2be01cf2c354354d99c2fc053025fab5","permalink":"https://oliviergimenez.github.io/blog/preprint_poaching/","publishdate":"2019-01-15T00:00:00Z","relpermalink":"/blog/preprint_poaching/","section":"blog","summary":"Check out our new @biorxivpreprint #preprint \u0026quot;Inferring #wildlife #poaching in Southeast Asia with multispecies dynamic #occupancy models\u0026quot; led by @LMarescot 🐯 https://t.co/g4NAQnUODw #HMMs @ArnaudLyet @rhohits #NeilCarter @SMARTCnsvTools @AgenceRecherche pic.twitter.com/5EttLyY1S9\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 15 janvier 2019 ","tags":["democom","carnivores","multispecies","occupancy models","poaching"],"title":"Inferring wildlife poaching in Southeast Asia with multispecies dynamic occupancy models","type":"blog"},{"authors":null,"categories":null,"content":"\u0026quot;Identifying conservation priorities for the critically endangered Balkanlynx\u0026quot; https://t.co/EDx0hUUT7v a nice article by #DimeMelovski on our recent @OryxTheJournal paper https://t.co/sVeTwbDlOc (15 days free access) w/ #rstat #OccupancyModels #GIS #maps https://t.co/Uv9uhsAPuY pic.twitter.com/c3iMLOChU3\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 14 janvier 2019kquote\u003e ","date":1547424000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1547424000,"objectID":"7b1362c1aa2b09fe41393d6816ba8d3d","permalink":"https://oliviergimenez.github.io/blog/comm_balkanlynx/","publishdate":"2019-01-14T00:00:00Z","relpermalink":"/blog/comm_balkanlynx/","section":"blog","summary":"\u0026quot;Identifying conservation priorities for the critically endangered Balkanlynx\u0026quot; https://t.co/EDx0hUUT7v a nice article by #DimeMelovski on our recent @OryxTheJournal paper https://t.co/sVeTwbDlOc (15 days free access) w/ #rstat #OccupancyModels #GIS #maps https://t.co/Uv9uhsAPuY pic.","tags":["large carnivores","Balkan lynx","occupancy models"],"title":"Identifying conservation priorities for the critically endangered Balkan lynx","type":"blog"},{"authors":null,"categories":null,"content":"In statistical ecology, we often need to calculate the sampling variance of a function of an estimate of which we do know the sampling variance. I keep forgetting how to implement the so-called delta method in R that allows to get an approximation of this quantity. So in this post I go through two examples in population ecology that should help me remembering. I use the deltamethod function from the msm package.\nLoad the package msm and get some help on the delta-method function:\nlibrary(msm) ?deltamethod Further examples can be obtained by typing in:\nexample(deltamethod) For a nice introduction to the delta method, check that out. Two papers are worth reading on the topic: \u0026lsquo;Who Invented the Delta Method?\u0026rsquo; by Jay M. Ver Hoef and \u0026lsquo;Approximating variance of demographic parameters using the delta method: A reference for avian biologists\u0026rsquo; by Larkin A. Powell.\nSimple example A simple example is when, for example, you get $\\phi$ (ignore the traditional hat) an estimate of a survival probability on the logit scale in some capture-recapture analyses, and you would like to get the standard error (SE) of survival on its natural scale.\nFor example, say $\\text{logit}(\\phi) = \\text{lphi} = -0.4473122$ with $\\text{SE} = 0.3362757$.\nTo obtain $\\phi$, you back-transform $\\text{lphi}$ using the reciprocal function of the logit function: $$\\phi = \\displaystyle{\\frac{\\exp(\\text{lphi})}{1+\\exp(\\text{lphi})}} = \\displaystyle{\\frac{1}{1+\\exp(\\text{-lphi})}} = \\displaystyle{\\frac{1}{1+\\exp(\\text{-(-0.4473122)})}} = 0.39.$$\nWhat about the SE of $\\phi$? Well, a direct application of the deltamethod function from the msm package gives the answer:\ndeltamethod(~ 1/(1+exp(-x1)), -0.4473122, 0.3362757^2) ## [1] 0.07999999 Two things to take care of:\nFirst, the variables in the formula must be labelled $x_1, x_2, \\text{text}$. You cannot use $x, y, z, \u0026hellip;$ for example. Just numbered $x$\u0026rsquo;s.\nSecond, the input parameters are the estimate and its squared SE (not the SE), and by default you will get as an output the SE (not the squared SE) of the function defined by the formula.\nComplex example This example deals with an occupancy model. It is a bit more complex than the previous example because we consider a function of several parameters for which we would like to calculate its SE. I assume that occupancy at first occasion was estimated along with its SE, and that one would like to obtain the SE of subsequent occupancy probabilities.\nI calculate time-dependent occupancy probabilities with the following formula $$\\psi_{t+1} = \\psi_t (1 - \\varepsilon) + (1 - \\psi_t) \\gamma$$ where $\\varepsilon$ is extinction, $\\gamma$ is colonisation and $\\psi_t$ is occupancy year $t$.\nWe assume that we obtained the following parameter estimates:\nepsilon = 0.39 gamma = 0.07 psi_init = 0.1 # first-occasion occupancy with corresponding SEs:\nse_epsilon = 0.08 se_psi_init = 0.01 se_gamma = 0.05 We will estimate occupancy and get SEs at 10 occasions, which we store in two matrices (column vectors):\npsi = matrix(0, nrow = 10, ncol = 1) psi_se = matrix(0, nrow = 10, ncol = 1) The first element is occupancy at first occasion:\npsi[1,] \u0026lt;- psi_init psi_se[1,] \u0026lt;- se_psi_init Then we iterate calculations using the formula above:\nfor(i in 2:10){ psi_current \u0026lt;- psi[i-1,] psi_se_current \u0026lt;- psi_se[i-1,] estmean \u0026lt;- c(psi_current,epsilon,gamma) estvar \u0026lt;- diag(c(psi_se_current,se_epsilon,se_gamma)^2) psi[i,] = (psi_current*(1-epsilon)) + ((1-psi_current)*gamma) # recurrence formula psi_se[i,] = deltamethod(~ x1*(1-x2) + (1-x1)*x3, estmean, estvar) # delta-method } Display results:\ndata.frame(psi = psi,sterr_psi = psi_se) ## psi sterr_psi ## 1 0.1000000 0.01000000 ## 2 0.1240000 0.04602347 ## 3 0.1369600 0.05132740 ## 4 0.1439584 0.05244394 ## 5 0.1477375 0.05259904 ## 6 0.1497783 0.05255782 ## 7 0.1508803 0.05250010 ## 8 0.1514753 0.05245886 ## 9 0.1517967 0.05243372 ## 10 0.1519702 0.05241934 Here, we assumed that sampling correlation was 0, in other words that the estimates of $\\psi$, $\\gamma$ and $\\epsilon$ were independent, hence the use of a diagonal matrix for estvar. It is possible to use a non-diagonal covariance matrix to account for non-null correlation.\n","date":1546560000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1546560000,"objectID":"7e0e44064021c26712608a491a4af7e7","permalink":"https://oliviergimenez.github.io/blog/delta-method/","publishdate":"2019-01-04T00:00:00Z","relpermalink":"/blog/delta-method/","section":"blog","summary":"In statistical ecology, we often need to calculate the sampling variance of a function of an estimate of which we do know the sampling variance. I keep forgetting how to implement the so-called delta method in R that allows to get an approximation of this quantity. So in this post I go through two examples in population ecology that should help me remembering. I use the deltamethod function from the msm package.\n","tags":["statistical ecology","estimation","delta-method","rstats"],"title":"Calculate the standard error of any function using the delta method","type":"blog"},{"authors":null,"categories":null,"content":"I read this awesome post (in French) by Baptiste Coulmont, professor in sociology, who explored the French academic network in sociology. Coulmont used the composition of PhD commitees to determine academic links between colleagues. The approach very appealing because it uses public data available from the website these.fr. Here, I used Coulmont\u0026rsquo;s R code to produce the French academic network in ecology. This was a nice opportunity to illustrate how to work in the tidyverse and to do some web scraping using the rvest package.\nGet the data Load the packages we need:\nlibrary(RCurl) library(tidyverse) library(lubridate) library(scales) library(hrbrthemes) library(data.table) # devtools::install_github(\u0026quot;privefl/bigreadr\u0026quot;) library(bigreadr) We now prepare the URL requests. The total number of PhDs is around 88000 on the period 2015-2018. Because the website uses slices of 1000 on each page, we proceed in sequence:\ni \u0026lt;- 1:88 i \u0026lt;- i*1000 URL \u0026lt;-paste0(\u0026quot;http://www.theses.fr/?q=\u0026amp;fq=dateSoutenance:[2015-01-01T23:59:59Z%2BTO%2B2018-12-31T23:59:59Z]\u0026amp;checkedfacets=\u0026amp;start=\u0026quot;,i,\u0026quot;\u0026amp;sort=none\u0026amp;status=\u0026amp;access=\u0026amp;prevision=\u0026amp;filtrepersonne=\u0026amp;zone1=titreRAs\u0026amp;val1=\u0026amp;op1=AND\u0026amp;zone2=auteurs\u0026amp;val2=\u0026amp;op2=AND\u0026amp;zone3=etabSoutenances\u0026amp;val3=\u0026amp;zone4=dateSoutenance\u0026amp;val4a=\u0026amp;val4b=\u0026amp;type=\u0026amp;lng=\u0026amp;checkedfacets=\u0026amp;format=csv\u0026quot;) Alternatively, the search can be done by hand directly from the theses.fr website. [François-Xavier Coudert] ( https://www.coudert.name/) also provides the search results for the 2015-2018 period.\nWe proceed with the requests, and store everything in a csv file:\nj \u0026lt;- 1 SERP \u0026lt;- 1 for(j in 1:length(URL)){ # loop over the slices SERP[j] \u0026lt;- getURL(URL[j]) write.csv(SERP,\u0026quot;SERP_2.csv\u0026quot;,append=F) } rm(SERP,i,j,URL) We keep only the PhDs in the field (Discipline) of ecology. This is basically the only change I have made to Coulmont\u0026rsquo;s neat code.\ntheses \u0026lt;- read.csv(\u0026quot;SERP_2.csv\u0026quot;,sep=\u0026quot;;\u0026quot;,quote=\u0026quot;\u0026quot;,skip=1,stringsAsFactors = F) #theses %\u0026gt;% # pull(X..Discipline..) %\u0026gt;% # unique() ecology \u0026lt;- theses %\u0026gt;% filter(grepl(\u0026quot;ecologie\u0026quot;,X..Discipline..,ignore.case=T)) %\u0026gt;% # keep PhDs with Displine == ecologie filter(X..Date.de.soutenance..!=\u0026quot;\u0026quot;) %\u0026gt;% # remove PhDs with missing dates of defense filter(X..Statut..==\u0026quot;soutenue\u0026quot;) # keep only PhDs that have been defended We now have the id of all PhDs in ecology defended during the period 2015-2018. We will use the id to get the composition of all PhD commitees. Getting this composition requires scraping the web page of each PhD, and to get the ID of each PhD. For doing so, we use the rvest package (see the excellent posts by Maëlle Salmon for examples).\nlibrary(rvest) identifiants \u0026lt;- ecology$X..Identifiant.de.la.these.. # get PhD ids reseau_total \u0026lt;- data_frame(noms_jury=\u0026quot;\u0026quot;, liens_jury=\u0026quot;\u0026quot;, these=\u0026quot;\u0026quot;, directeurs=\u0026quot;\u0026quot;, liens_directeurs=\u0026quot;\u0026quot;) for (i in 1:length(identifiants)) { # get info on current PhD data_theses_eco \u0026lt;- read_html( paste0(\u0026quot;http://www.theses.fr/\u0026quot;,identifiants[i]) ) # get name PhD supervisor for directeurs \u0026lt;- bind_cols( directeurs = data_theses_eco %\u0026gt;% html_nodes(\u0026quot;div .donnees-ombre p\u0026quot;) %\u0026gt;% .[[1]] %\u0026gt;% html_nodes(\u0026quot;a\u0026quot;) %\u0026gt;% html_text() , liens_directeurs = data_theses_eco %\u0026gt;% html_nodes(\u0026quot;div .donnees-ombre p\u0026quot;) %\u0026gt;% .[[1]] %\u0026gt;% html_nodes(\u0026quot;a\u0026quot;) %\u0026gt;% html_attr(name=\u0026quot;href\u0026quot;) ) %\u0026gt;% mutate( these = identifiants[i] ) # get names of people in commitees jury \u0026lt;- bind_cols( noms_jury = data_theses_eco %\u0026gt;% html_nodes(\u0026quot;div .donnees p a\u0026quot;) %\u0026gt;% html_text() , liens_jury = data_theses_eco %\u0026gt;% html_nodes(\u0026quot;div .donnees p a\u0026quot;) %\u0026gt;% html_attr(name=\u0026quot;href\u0026quot;) ) %\u0026gt;% mutate( these = identifiants[i] ) # put all together reseau \u0026lt;- jury %\u0026gt;% left_join(directeurs,by=\u0026quot;these\u0026quot;) reseau_total \u0026lt;- bind_rows(reseau_total,reseau) } Build the network Load the packages we need, and the data we got at the previous step:\nlibrary(igraph) library(ggraph) library(ggrepel) load('reseau_total.RData') Coulmont defined a weighted link between two colleagues i and j as follows: 3 if i and j are both supervisors, 2 if i is a supervisor and j a PhD commitee member and 1 if both i and j are PhD commitee members. A colleague may accumulate several weights.\ndirections_theses \u0026lt;- reseau_total %\u0026gt;% select(these,directeurs) %\u0026gt;% unique() %\u0026gt;% group_by(these) %\u0026gt;% mutate(N=n()) %\u0026gt;% filter(N==2) %\u0026gt;% # keep co-supervision w/ 2 supervisors mutate(rang=rank(directeurs)) %\u0026gt;% spread(key=rang,value=directeurs) %\u0026gt;% ungroup() %\u0026gt;% select(nom1=`1`,nom2=`2`) %\u0026gt;% mutate(poids=3) directions_jury \u0026lt;- reseau_total %\u0026gt;% select(nom1=noms_jury,nom2=directeurs) %\u0026gt;% filter( nom1 != \u0026quot;\u0026quot;) %\u0026gt;% mutate(poids=2) %\u0026gt;% group_by(nom1,nom2) %\u0026gt;% summarize(poids=sum(poids)) jury_jury \u0026lt;- reseau_total %\u0026gt;% select(noms_jury,these) %\u0026gt;% unique() %\u0026gt;% filter(noms_jury!=\u0026quot;\u0026quot;) g_j \u0026lt;- graph_from_data_frame(jury_jury,directed=F) V(g_j)$type \u0026lt;- V(g_j)$name %in% jury_jury$noms_jury g_j_1 \u0026lt;- bipartite_projection(g_j,which=\u0026quot;true\u0026quot;) jurys \u0026lt;- as_long_data_frame(g_j_1) %\u0026gt;% select(nom1=`ver[el[, 1], ]`, nom2=`ver2[el[, 2], ]`, poids=weight) reseau_petit \u0026lt;- bind_rows(directions_theses,directions_jury,jurys) %\u0026gt;% group_by(nom1,nom2) %\u0026gt;% summarize(poids=sum(poids)) # data.frame from which the network will be created Each node in the network has a size proportional to its betweenness score. We also determine communities using the walktrap algorithm that will be colored differently. The width of an edge is proportional to the strength of the link between the two corresponding nodes.\ng \u0026lt;- graph_from_data_frame(reseau_petit, directed=F) # create network from data.frame g \u0026lt;- simplify(g,edge.attr.comb = sum) V(g)$degres \u0026lt;- degree(g) V(g)$label \u0026lt;- gsub(\u0026quot;^\\\\S+\\\\s+(.+)$\u0026quot;,\u0026quot;\\\\1\u0026quot;,V(g)$name) V(g)$communaute \u0026lt;- as.character(cluster_walktrap(g, steps=15)$membership) # determine communities V(g)$closeness \u0026lt;- (5*closeness(g))^10 V(g)$btwns \u0026lt;- betweenness(g) # network metric betweeness V(g)$eigen_centr \u0026lt;- eigen_centrality(g)$vector g \u0026lt;- delete_edges(g, which(E(g)$poids\u0026lt;5) ) # delete edges with weight \u0026lt;= 4 V(g)$cluster_number \u0026lt;- clusters(g)$membership # to which community you belong g \u0026lt;- induced_subgraph(g, V(g)$cluster_number== which( max(clusters(g)$csize) == clusters(g)$csize) ) E(g)$weight \u0026lt;- 1/E(g)$poids # width of edge proportional to weight V(g)$label \u0026lt;- ifelse(V(g)$degres\u0026lt;20,\u0026quot;\u0026quot;,V(g)$label) # do not display all names Plot the network We now plot the network. For clarity, we only indicate the names of colleagues who were part of several phD commitees.\nggraph(g,layout=\u0026quot;igraph\u0026quot;,algorithm=\u0026quot;fr\u0026quot;) + geom_edge_link(aes(width=.1*poids), alpha=.1, end_cap = circle(5, 'mm'), start_cap = circle(5, 'mm')) + geom_node_point(aes(size=eigen_centr), color=\u0026quot;white\u0026quot;,alpha=1) + geom_node_point(aes(color=communaute,size=eigen_centr), alpha=.5) + scale_size_area(max_size = 20) + geom_node_text(aes(label=label),size=3,repel=T,box.padding = 0.15) + labs(title=\u0026quot;Réseaux des écologues\u0026quot;, subtitle=\u0026quot;Soutenances de thèses entre 2015 et 2018\u0026quot;, caption=\u0026quot;Sources : theses.fr \\n Code par B. Coulmont, modifié par O. Gimenez\u0026quot;) + theme_graph(foreground = 'white', fg_text_colour = 'white', base_family = \u0026quot;Helvetica\u0026quot;) + theme(legend.position=\u0026quot;none\u0026quot;, text=element_text(size=16,family=\u0026quot;Helvetica\u0026quot;), plot.margin = unit(c(0.2, 0.2, 0.2, 0.2), units=\u0026quot;line\u0026quot;)) # save ggsave(filename = \u0026quot;ecology_network.pdf\u0026quot;,width=30,height = 20) I played around the defaults Coulmont used to build and plot the network. It helps in getting a better understanding of the network and the links between colleagues working in ecology. Overall, I indeed feel very much connected to my colleagues in Montpellier, Lyon and Grenoble. I should probably go out of my comfort zone and interact even more with my colleagues from La Rochelle, Marseille and Aix-en-Provence 😃\nAs always, data and code are available from GitHub.\n","date":1546387200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1546387200,"objectID":"8cc6325e87401f224bee7aedb8164272","permalink":"https://oliviergimenez.github.io/blog/network_ecology/","publishdate":"2019-01-02T00:00:00Z","relpermalink":"/blog/network_ecology/","section":"blog","summary":"I read this awesome post (in French) by Baptiste Coulmont, professor in sociology, who explored the French academic network in sociology. Coulmont used the composition of PhD commitees to determine academic links between colleagues. The approach very appealing because it uses public data available from the website these.fr. Here, I used Coulmont\u0026rsquo;s R code to produce the French academic network in ecology. This was a nice opportunity to illustrate how to work in the tidyverse and to do some web scraping using the rvest package.\n","tags":["network","ecology","tidyverse","rstats"],"title":"Scientific research is all about networking","type":"blog"},{"authors":null,"categories":null,"content":"Our team has had a fantastic 2018 year! Inspired by @Maestrelab, below is a thread of some highlights 🍾😜🤩 @CNRS_OccitaniE @INEE_CNRS @umontpellier @IsiteMUSE @LMarescot @NSantostasi @MaudQueroue @ValentinLauret @oksanagrente @GueryLorelei @roques_seb https://t.co/FaP9o0bOD4.\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 23 décembre 2018 \u0026lt;\nblockquote class=\u0026ldquo;twitter-tweet\u0026rdquo; data-lang=\u0026ldquo;fr\u0026rdquo;\u0026gt;1. We published papers in #statistalecology \u0026amp; large mammals #conservation w/ interest in human \u0026amp; social sciences https://t.co/Xlh6reUUQc @Oikos_Journal @PopulEcology @AnimalEcology @JOSS_TheOJ @OryxTheJournal @EcographyJourna @CommsBio @FunEcology @MethodsEcolEvol @ConBiology. pic.twitter.com/FvakGRzacx\n— Olivier Gimenez 💤 (@oaggimenez) 23 décembre 2018\n2. We had amazing interns! D. Folio analysed maternal traits \u0026amp; polar bears reproduction w/ @SCubi25 \u0026amp; J. Aars @NorskPolar; A. Hendrick studied ethnology \u0026amp; whale-whatching w/ N. Lescureux \u0026amp; H. Labach @GIS_3M; @ValentinLauret developed adaptive monitoring for a #MPAs network. pic.twitter.com/7RCc9hAXgm\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 23 décembre 2018 3. Gilles Maurer and @LouvrierJulie successfully defended their PhD on the the interactions between captive \u0026amp; wild pops of 🐘 w/ @zoobeauval and the statistical #modelling of large #carnivores 🐺🐱 in Europe w/ @ONCFS_officiel \u0026amp; @umontpellier @CollegeDoctoUM. pic.twitter.com/rAtJifrTb6\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 23 décembre 2018 4. We engaged w/ stakeholders to try and have an impact on wildlife management. Several projects are on tracks in that direction https://t.co/R4UcwpDxVN. pic.twitter.com/xovnHc6HCq\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 23 décembre 2018 5. We had awesome visitors (remotely or not) who contributed to the #boulet spirit. Looking in your direction @AmandaVincent1, @u_arbieu, C. Oosthuizen, @pardo_deborah, @saveourseas, @StatnMap, @FredBarraquand, @MoonbeamLevels, @BenhaiemSarah, @CROC_Estelle and others I forget. pic.twitter.com/nF12RjpZrs\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 23 décembre 2018 6. New adventures for 2019 include embarking in 3 new PhDs: @ValentinLauret on 🐬 #adaptive #monitoring in #MPAs w/ @GIS_3M, @MaudQueroue on multispecies 🐦 #demography w/ #CEBC @UMR_MECADEV and @oksanagrente on 🐺 #adaptive #management w/ @ONCFS_officiel @chamaillejammes. pic.twitter.com/nCGfaruLIf\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 23 décembre 2018 7. We also spent time planning the future, and I can tell you, 2019 is gonna be even more awesome than 2018 w/ our new team on the interactions between humans and wildlife #interdisciplinarity #management #shs @abesnardEPHE @INEE_CNRS @inshs_cnrs @CNRS_OccitaniE. pic.twitter.com/3TEAUzC5Ho\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 23 décembre 2018 We https://t.co/FaP9o0bOD4 wish you Merry Christmas and Happy New Year, all the best and see you in 2019 🎄🤶🎅\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 23 décembre 2018 ","date":1546041600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1546041600,"objectID":"39ffc16e5d0f63244c2906d93acb5d62","permalink":"https://oliviergimenez.github.io/blog/2018retrospective/","publishdate":"2018-12-29T00:00:00Z","relpermalink":"/blog/2018retrospective/","section":"blog","summary":"Our team has had a fantastic 2018 year! Inspired by @Maestrelab, below is a thread of some highlights 🍾😜🤩 @CNRS_OccitaniE @INEE_CNRS @umontpellier @IsiteMUSE @LMarescot @NSantostasi @MaudQueroue @ValentinLauret @oksanagrente @GueryLorelei @roques_seb https://t.co/FaP9o0bOD4.\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 23 décembre 2018 \u0026lt;\n","tags":["retrospective","boulet team"],"title":"Looking back at 2018","type":"blog"},{"authors":null,"categories":null,"content":"Pretty proud of my modest contribution: data and #rstats code https://t.co/EUG4IgPdKR 🤓\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 3 décembre 2018 Out now: The Balkan #lynx is one of the rarest, most threatened \u0026amp; least-studied large carnivores. Melovski et al. used questionnaire surveys \u0026amp; occupancy modelling to identify conservation priorities for this Critically Endangered subspecies.https://t.co/asollm8Ior pic.twitter.com/af6jTFd9EI\n\u0026mdash; Oryx (@OryxTheJournal) 5 décembre 2018 ","date":1543795200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1543795200,"objectID":"09dd38e35b5f5080d9bbdf4dd66b155d","permalink":"https://oliviergimenez.github.io/blog/articlebalkanlynx/","publishdate":"2018-12-03T00:00:00Z","relpermalink":"/blog/articlebalkanlynx/","section":"blog","summary":"Pretty proud of my modest contribution: data and #rstats code https://t.co/EUG4IgPdKR 🤓\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 3 décembre 2018 Out now: The Balkan #lynx is one of the rarest, most threatened \u0026amp; least-studied large carnivores.","tags":["balkan lynx","occupancy models","conservation"],"title":"New paper on the Balkan lynx","type":"blog"},{"authors":null,"categories":null,"content":"Check out our @CommsBio paper w/ @IZWberlin \u0026quot;Slow recovery from a disease #epidemic in the #spotted #hyena, a keystone social #carnivore\u0026quot; led by #SarahBenhaiem \u0026amp; @LMarescot @CNRS_OccitaniE @INEE_CNRS @IsiteMUSE @umontpellier #matrixmodel #capturerecapture pic.twitter.com/kangaJYtha\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 212018 ","date":1542758400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1542758400,"objectID":"b2f5501d98dfad3551832c922ee4d6ce","permalink":"https://oliviergimenez.github.io/blog/articlesarahetlulu/","publishdate":"2018-11-21T00:00:00Z","relpermalink":"/blog/articlesarahetlulu/","section":"blog","summary":"Check out our @CommsBio paper w/ @IZWberlin \u0026quot;Slow recovery from a disease #epidemic in the #spotted #hyena, a keystone social #carnivore\u0026quot; led by #SarahBenhaiem \u0026amp; @LMarescot @CNRS_OccitaniE @INEE_CNRS @IsiteMUSE @umontpellier #matrixmodel #capturerecapture pic.","tags":["hyenas","matrix models","disease ecology"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"Thanks for your visit @u_arbieu, the great talk and the stimulating discussions! Awesome collaborations to come 🤝🍻 🇩🇪🇫🇷🇮🇹🐺 https://t.co/04goJ5iMAI @NSantostasi @LouvrierJulie @roques_seb @oksanagrente @MaudQueroue https://t.co/exZsi2qxOA\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 16ovembre 2018 ","date":1542326400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1542326400,"objectID":"d8f6f87482f258a5f6a6af600659f58d","permalink":"https://oliviergimenez.github.io/blog/ugoarbieu_visit/","publishdate":"2018-11-16T00:00:00Z","relpermalink":"/blog/ugoarbieu_visit/","section":"blog","summary":"Thanks for your visit @u_arbieu, the great talk and the stimulating discussions! Awesome collaborations to come 🤝🍻 🇩🇪🇫🇷🇮🇹🐺 https://t.co/04goJ5iMAI @NSantostasi @LouvrierJulie @roques_seb @oksanagrente @MaudQueroue https://t.co/exZsi2qxOA\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 16ovembre 2018 ","tags":["large carnivores","wolf","human-wildlife conflicts"],"title":"Ugo Arbieu's visit","type":"blog"},{"authors":null,"categories":null,"content":"🐬🇬🇷🇮🇹🇫🇷 New paper by @NSantostasi @INEE_CNRS @CNRS_OccitaniE @IsiteMUSE @umontpellier 🤩👏 https://t.co/6vQ6d9HevV\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 9 novembre 2018 ","date":1541721600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1541721600,"objectID":"cd47ba865110597cd03cd99fcde68978","permalink":"https://oliviergimenez.github.io/blog/articlenina/","publishdate":"2018-11-09T00:00:00Z","relpermalink":"/blog/articlenina/","section":"blog","summary":"🐬🇬🇷🇮🇹🇫🇷 New paper by @NSantostasi @INEE_CNRS @CNRS_OccitaniE @IsiteMUSE @umontpellier 🤩👏 https://t.co/6vQ6d9HevV\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 9 novembre 2018 ","tags":["dolphin","R","rstats","PVA","Greece"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"#INLA workshop on spatio-temporal models in the beautiful city of #Avignon 🤩 #RESSTE #GdREcoStat @oksanagrente @CREEM_cake pic.twitter.com/nuLhIkMiwO\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 718 ","date":1541462400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1541462400,"objectID":"cdb3ab92fd571bd8b2dec3589c64083d","permalink":"https://oliviergimenez.github.io/blog/inla_workshop/","publishdate":"2018-11-06T00:00:00Z","relpermalink":"/blog/inla_workshop/","section":"blog","summary":"#INLA workshop on spatio-temporal models in the beautiful city of #Avignon 🤩 #RESSTE #GdREcoStat @oksanagrente @CREEM_cake pic.twitter.com/nuLhIkMiwO\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 718 ","tags":["INLA","R","rstats","spatio-temporal models","point process"],"title":"Workshop on spatio-temporal models with INLA","type":"blog"},{"authors":null,"categories":null,"content":"The famous #NetLogo butterfly example coded in #rstats w/ #NetLogoR pic.twitter.com/VbUUa5vIep\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 5 novembre 2018 Wolf model from Marucco \u0026amp; @eliotmcintire coded in #rstats w/ #NetLogoR \u0026amp; #SpaDES https://t.co/ERx71CIgZv pic.twitter.com/f0YTHuKGSS\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 5 novembre 2018 ","date":1541376000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1541376000,"objectID":"c63044ebe62464f697477c23a120aa89","permalink":"https://oliviergimenez.github.io/blog/crashcourse_netlogor/","publishdate":"2018-11-05T00:00:00Z","relpermalink":"/blog/crashcourse_netlogor/","section":"blog","summary":"The famous #NetLogo butterfly example coded in #rstats w/ #NetLogoR pic.twitter.com/VbUUa5vIep\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 5 novembre 2018 Wolf model from Marucco \u0026amp; @eliotmcintire coded in #rstats w/ #NetLogoR \u0026amp; #SpaDES https://t.","tags":["IBM","R","rstats","modelling"],"title":"Crash course on individual-based models using NetLogoR","type":"blog"},{"authors":null,"categories":null,"content":"🍃📊 La @sfecologie créé avec le #GDR #ecostat un groupe thématique #écologie #statistique https://t.co/Lv33Ur1POc - site du GDR https://t.co/kTXpjJXlya \u0026amp; liste diffusion https://t.co/y3s0xgyzEJ - plus sur l\u0026#39;écologie statistique https://t.co/aOvTJFGDLu \u0026amp; https://t.co/7TRCKNUMKi pic.twitter.com/C4ifYZ48QC\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 4 novembre 2018 ","date":1541203200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1541203200,"objectID":"4b6e42cd605903d609a47e00ced9022e","permalink":"https://oliviergimenez.github.io/blog/groupeecostatsfee/","publishdate":"2018-11-03T00:00:00Z","relpermalink":"/blog/groupeecostatsfee/","section":"blog","summary":"🍃📊 La @sfecologie créé avec le #GDR #ecostat un groupe thématique #écologie #statistique https://t.co/Lv33Ur1POc - site du GDR https://t.co/kTXpjJXlya \u0026amp; liste diffusion https://t.co/y3s0xgyzEJ - plus sur l\u0026#39;écologie statistique https://t.co/aOvTJFGDLu \u0026amp; https://t.","tags":["book","write"],"title":"Groupe thématique Ecologie Statistique de la SFEE","type":"blog"},{"authors":null,"categories":null,"content":"🐘🌏 Sauver l\u0026#39;éléphant d\u0026#39;Asie avec des congés maternités? Itw de Gilles Maurer dans le magazine #LUM de l\u0026#39;@umontpellier https://t.co/sKEJc8FsJH @IsiteMUSE @CNRSenLR @INEE_CNRS @zoobeauval @UWyonews - article scientifique https://t.co/uqqPfPDvpp et 🎦 https://t.co/1m9YAoe3Q3 pic.twitter.com/xxMxFjtrKS\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 4 novembre 2018 ","date":1541030400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1541030400,"objectID":"164a183433e0151de3144fb916a5fac3","permalink":"https://oliviergimenez.github.io/blog/howtowritealot/","publishdate":"2018-11-01T00:00:00Z","relpermalink":"/blog/howtowritealot/","section":"blog","summary":"🐘🌏 Sauver l\u0026#39;éléphant d\u0026#39;Asie avec des congés maternités? Itw de Gilles Maurer dans le magazine #LUM de l\u0026#39;@umontpellier https://t.co/sKEJc8FsJH @IsiteMUSE @CNRSenLR @INEE_CNRS @zoobeauval @UWyonews - article scientifique https://t.co/uqqPfPDvpp et 🎦 https://t.","tags":["vulgarisation","Asian elephant"],"title":"Sauver l'éléphant d'Asie avec des congés maternités?","type":"blog"},{"authors":null,"categories":null,"content":"Nice read \u0026quot;How to write a lot\u0026quot; by Paul Silvia: \u0026quot;Make a writing schedule and show up for it. Want less and do more\u0026quot;, \u0026quot;Productive writing involves harnessing the power of habit, and habits come from repetition\u0026quot; Let\u0026#39;s get to it then! https://t.co/TTAOrFl6Z5 https://t.co/KXgGCq9R2v pic.twitter.com/Eno5qv3g0w\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 2 novembre 201 ","date":1540857600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1540857600,"objectID":"8a285862acdea500036f71167b3da327","permalink":"https://oliviergimenez.github.io/blog/articlelumgilles/","publishdate":"2018-10-30T00:00:00Z","relpermalink":"/blog/articlelumgilles/","section":"blog","summary":"Nice read \u0026quot;How to write a lot\u0026quot; by Paul Silvia: \u0026quot;Make a writing schedule and show up for it. Want less and do more\u0026quot;, \u0026quot;Productive writing involves harnessing the power of habit, and habits come from repetition\u0026quot; Let\u0026#39;s get to it then!","tags":["book","write"],"title":"How to write a lot?","type":"blog"},{"authors":null,"categories":null,"content":"Check out our new preprint by awesome Anne-Sophie Bonnet-Lebrun \u0026quot;Identifying priority conservation areas for recovering large #carnivores using #citizenscience data\u0026quot; w/ 🐟 point-process models to map 🇬🇷 🐻 distribution https://t.co/CEUbVKuXRQ starring @Arcturosgr @MoonbeamLevels pic.twitter.com/fb9UAP3Aep\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 29 oct8 ","date":1540771200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1540771200,"objectID":"a96ec3dd3f50f67a57a165ab894d5345","permalink":"https://oliviergimenez.github.io/blog/newpreprint/","publishdate":"2018-10-29T00:00:00Z","relpermalink":"/blog/newpreprint/","section":"blog","summary":"Check out our new preprint by awesome Anne-Sophie Bonnet-Lebrun \u0026quot;Identifying priority conservation areas for recovering large #carnivores using #citizenscience data\u0026quot; w/ 🐟 point-process models to map 🇬🇷 🐻 distribution https://t.co/CEUbVKuXRQ starring @Arcturosgr @MoonbeamLevels pic.","tags":["sdm","large carnivores","brown bear","point process"],"title":"New preprint!","type":"blog"},{"authors":null,"categories":null,"content":"New interdisciplinary project combining ecological, human and social sciences for large #carnivores management 🐺🐻🐱 https://t.co/04goJ5iMAI @INEE_CNRS @inshs_cnrs @CNRSenLR @ONCFS_officiel @univpaulvalery @CNRS @PACTE_grenoble pic.twitter.com/7IVzJR7l8y\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 17 octobre 2018 Nouveau projet interdisciplinaire combinant sciences écologiques, humaines et sociales pour la gestion des grands #carnivores 🐺🐻🐱 https://t.co/04goJ5iMAI @INEE_CNRS @inshs_cnrs @CNRSenLR @ONCFS_officiel @univpaulvalery @CNRS @PACTE_grenoble pic.twitter.com/BP8rVPkvtD\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 16 ocobre 2018 ","date":1538956800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1538956800,"objectID":"91211eecddc39b58dcaa7a64c69a4089","permalink":"https://oliviergimenez.github.io/blog/ecosocar_website/","publishdate":"2018-10-08T00:00:00Z","relpermalink":"/blog/ecosocar_website/","section":"blog","summary":"New interdisciplinary project combining ecological, human and social sciences for large #carnivores management 🐺🐻🐱 https://t.co/04goJ5iMAI @INEE_CNRS @inshs_cnrs @CNRSenLR @ONCFS_officiel @univpaulvalery @CNRS @PACTE_grenoble pic.twitter.com/7IVzJR7l8y\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 17 octobre 2018 Nouveau projet interdisciplinaire combinant sciences écologiques, humaines et sociales pour la gestion des grands #carnivores 🐺🐻🐱 https://t.","tags":["interdisciplinarity","large carnivores","ecosocar"],"title":"New project!","type":"blog"},{"authors":null,"categories":null,"content":"Meeting about the projet of our new team on the ecology of interactions between humans and wildlife #interdisciplinarity #modelling #management #geography #ethnology #shs #teamboulet @umontpellier @IsiteMUSE @CNRSenLR @INEE_CNRS @inshs_cnrs pic.twitter.com/Vj3OqIkFr6\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 8 tobre 2018 ","date":1538956800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1538956800,"objectID":"b90e244256159d659716777d0c8f9508","permalink":"https://oliviergimenez.github.io/blog/newteam/","publishdate":"2018-10-08T00:00:00Z","relpermalink":"/blog/newteam/","section":"blog","summary":"Meeting about the projet of our new team on the ecology of interactions between humans and wildlife #interdisciplinarity #modelling #management #geography #ethnology #shs #teamboulet @umontpellier @IsiteMUSE @CNRSenLR @INEE_CNRS @inshs_cnrs pic.twitter.com/Vj3OqIkFr6","tags":["interdisciplinarity","team"],"title":"Team meeting","type":"blog"},{"authors":null,"categories":null,"content":"We had fun 😉 pic.twitter.com/C93lSMzts8\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 7octobre 2018 ","date":1538524800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1538524800,"objectID":"91a2ffd4067944ed9c7fb602085bf1e1","permalink":"https://oliviergimenez.github.io/blog/4thmeetinggdrecostat/","publishdate":"2018-10-03T00:00:00Z","relpermalink":"/blog/4thmeetinggdrecostat/","section":"blog","summary":"We had fun 😉 pic.twitter.com/C93lSMzts8\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 7octobre 2018 ","tags":["statistical ecology","GDR"],"title":"4th meeting of the GDR EcoStat","type":"blog"},{"authors":null,"categories":null,"content":"More details here https://t.co/FaP9o0bOD4\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 1 octobre 2018 ","date":1538352000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1538352000,"objectID":"16d33562bff7277d69b5ad6feb653e60","permalink":"https://oliviergimenez.github.io/blog/new_phds/","publishdate":"2018-10-01T00:00:00Z","relpermalink":"/blog/new_phds/","section":"blog","summary":"More details here https://t.co/FaP9o0bOD4\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 1 octobre 2018 ","tags":["phd","boulet"],"title":"Lucky to have 3 new PhD students","type":"blog"},{"authors":null,"categories":null,"content":"🤩🍾🥳 Well done @LouvrierJulie - Impressiiiiiiiiiiiiive work! https://t.co/wd99KZcYTy\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 30 septembre 2018 ","date":1538265600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1538265600,"objectID":"b777688efbb0c5644da2f06511a79dc7","permalink":"https://oliviergimenez.github.io/blog/phdjulie_delivered/","publishdate":"2018-09-30T00:00:00Z","relpermalink":"/blog/phdjulie_delivered/","section":"blog","summary":"🤩🍾🥳 Well done @LouvrierJulie - Impressiiiiiiiiiiiiive work! https://t.co/wd99KZcYTy\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 30 septembre 2018 ","tags":["large carnivores","sdm","phd","boulet","wolf","lynx","occupancy"],"title":"Julie Louvrier submitted her PhD manuscript!","type":"blog"},{"authors":null,"categories":null,"content":"🐬👩 💻 New paper by awesome @NSantostasi showing that \u0026quot;Common dolphins in the Gulf of Corinth are Critically Endangered\u0026quot; @IUCNRedList 🇬🇷🇮🇹🇫🇷 #pva #conservation @marinemammalogy https://t.co/lcQpyoetm8 #womeninSTEM pic.twitter.com/21raWOutfM\n\u0026mdash; Olivier Gimenez 😴 (@oaggimenez) 30 septembre 2018 ","date":1538265600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1538265600,"objectID":"c7dfbd5504a27ae4697c62d7b37a9a1e","permalink":"https://oliviergimenez.github.io/blog/nina_pva_commondolphin/","publishdate":"2018-09-30T00:00:00Z","relpermalink":"/blog/nina_pva_commondolphin/","section":"blog","summary":"🐬👩 💻 New paper by awesome @NSantostasi showing that \u0026quot;Common dolphins in the Gulf of Corinth are Critically Endangered\u0026quot; @IUCNRedList 🇬🇷🇮🇹🇫🇷 #pva #conservation @marinemammalogy https://t.co/lcQpyoetm8 #womeninSTEM pic.twitter.com/21raWOutfM\n\u0026mdash; Olivier Gimenez 😴 (@oaggimenez) 30 septembre 2018 ","tags":["common dolphins","pva","conservation","cetaceans","dolphin biology and conservation","Greece"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"🤩🐺 New paper by unstoppable @LouvrierJulie \u0026quot;Accounting for #misidentification and #heterogeneity in #occupancy studies using #HMMs\u0026quot; https://t.co/UqJfbkFjbW #teamboulet #openaccess @EcoHMM #carnivorepapersbywomen @ONCFS_officiel @umontpellier @IsiteMUSE @CNRSenLR @INEE_CNRS pic.twitter.com/YGgPNjzRUE\n\u0026mdash; Olivier Gimenez 🥐 (@oaggimenez) 19 septembre18 ","date":1537315200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1537315200,"objectID":"dc432ba16181b9989c1f73e044ff9f6b","permalink":"https://oliviergimenez.github.io/blog/julie_ecolmodel_paper/","publishdate":"2018-09-19T00:00:00Z","relpermalink":"/blog/julie_ecolmodel_paper/","section":"blog","summary":"🤩🐺 New paper by unstoppable @LouvrierJulie \u0026quot;Accounting for #misidentification and #heterogeneity in #occupancy studies using #HMMs\u0026quot; https://t.co/UqJfbkFjbW #teamboulet #openaccess @EcoHMM #carnivorepapersbywomen @ONCFS_officiel @umontpellier @IsiteMUSE @CNRSenLR @INEE_CNRS pic.twitter.com/YGgPNjzRUE\n\u0026mdash; Olivier Gimenez 🥐 (@oaggimenez) 19 septembre18 ","tags":["HMM","occupancy","misidentification","false positives","carnivores"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"New paper by #SarahBenhaiem \u0026amp; @LMarescot 🤩👏 \u0026quot;Robustness of Eco-Epidemiological Capture-Recapture Parameter Estimates to Variation in Infection State Uncertainty\u0026quot; https://t.co/ZvNghUHUUY #HMM #multievent #TMB #WomenInSTEM #carnivorepapersbywomen #heynas #serengeti pic.twitter.com/VT6iStk5UW\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 28 août 2018 ","date":1535414400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1535414400,"objectID":"e8bddb7f4ad792cd14721a438c39f725","permalink":"https://oliviergimenez.github.io/blog/sarahlulu_paper/","publishdate":"2018-08-28T00:00:00Z","relpermalink":"/blog/sarahlulu_paper/","section":"blog","summary":"New paper by #SarahBenhaiem \u0026amp; @LMarescot 🤩👏 \u0026quot;Robustness of Eco-Epidemiological Capture-Recapture Parameter Estimates to Variation in Infection State Uncertainty\u0026quot; https://t.co/ZvNghUHUUY #HMM #multievent #TMB #WomenInSTEM #carnivorepapersbywomen #heynas #serengeti pic.twitter.com/VT6iStk5UW\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 28 août 2018 ","tags":["multievent","capture-recapture","TMB","disease ecology","carnivores"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"","date":1534032000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1534032000,"objectID":"945017ff221838829cc693a7918149ae","permalink":"https://oliviergimenez.github.io/my-project/external-project/sciencepolicy/","publishdate":"2018-08-12T00:00:00Z","relpermalink":"/my-project/external-project/sciencepolicy/","section":"my-project","summary":"Combining ecological, human and social sciences for large carnivores management.","tags":["carnivores","ecology","social sciences","policy"],"title":"Socio-ecological management of large carnivores","type":"my-project"},{"authors":null,"categories":null,"content":"\u0026quot;Tenir compte de la dimension humaine dans la gestion des #conflits hommes-grands #carnivores : le cas de l’#ours brun dans les #Pyrenees\u0026quot; https://t.co/OJYC7Lje9L paru dans Faune Sauvage @ONCFS_officiel reprenant nos résultats publiés dans Biol. Cons. https://t.co/UgPsN37NSA pic.twitter.com/oRU7kLyd7Y\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 1 août 2018 ","date":1532217600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1532217600,"objectID":"70326c31b08f1284838e43542839e5ac","permalink":"https://oliviergimenez.github.io/blog/ours_faunesauvage/","publishdate":"2018-07-22T00:00:00Z","relpermalink":"/blog/ours_faunesauvage/","section":"blog","summary":"\u0026quot;Tenir compte de la dimension humaine dans la gestion des #conflits hommes-grands #carnivores : le cas de l’#ours brun dans les #Pyrenees\u0026quot; https://t.co/OJYC7Lje9L paru dans Faune Sauvage @ONCFS_officiel reprenant nos résultats publiés dans Biol.","tags":["scicomm","bear","social sciences"],"title":"Sciences sociales et ours brun dans les Pyrénées","type":"blog"},{"authors":null,"categories":null,"content":"🤩🐱 New paper by unstoppable @LouvrierJulie \u0026quot;Use of ambiguous detections to improve estimates from #SDMs\u0026quot; in @ConBiology https://t.co/XXZFGDT7ko #teamboulet @ZiFridolin @aveamphibious #WomenInSTEM #carnivorepapersbywomen @ONCFS_officiel @umontpellier @IsiteMUSE @CNRSenLR pic.twitter.com/LkZtbswikJ\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 20 juillet8 ","date":1532044800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1532044800,"objectID":"c706981fd52e0822c866f41be0f58617","permalink":"https://oliviergimenez.github.io/blog/julie_paper/","publishdate":"2018-07-20T00:00:00Z","relpermalink":"/blog/julie_paper/","section":"blog","summary":"🤩🐱 New paper by unstoppable @LouvrierJulie \u0026quot;Use of ambiguous detections to improve estimates from #SDMs\u0026quot; in @ConBiology https://t.co/XXZFGDT7ko #teamboulet @ZiFridolin @aveamphibious #WomenInSTEM #carnivorepapersbywomen @ONCFS_officiel @umontpellier @IsiteMUSE @CNRSenLR pic.twitter.com/LkZtbswikJ\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 20 juillet8 ","tags":["lynx","occupancy","misendification"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"Following @JamesGrecian\u0026#39;s tracks (who helped me to get the tweets), I did some analyses of the #isec2018 tweets largely inspired (if not copy and pasted) by previous analyses from @brenborbon, @_ColinFay and @bryangoodrich - Data \u0026amp; #rstats codes https://t.co/ZCv4DAeS5S pic.twitter.com/YmU2QBsyp2\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 9 juillt 2018 ","date":1531094400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1531094400,"objectID":"cc0ff6f61ddd2442474580f25fadc1aa","permalink":"https://oliviergimenez.github.io/blog/tweetisec2018/","publishdate":"2018-07-09T00:00:00Z","relpermalink":"/blog/tweetisec2018/","section":"blog","summary":"Following @JamesGrecian\u0026#39;s tracks (who helped me to get the tweets), I did some analyses of the #isec2018 tweets largely inspired (if not copy and pasted) by previous analyses from @brenborbon, @_ColinFay and @bryangoodrich - Data \u0026amp; #rstats codes https://t.","tags":["isec","statistical ecology","twitter"],"title":"ISEC 2018 tweets analysis","type":"blog"},{"authors":null,"categories":null,"content":"And nooooooow, ladies and gentlemen, @LMarescot presents her work on multispecies occupancy models to infer poaching #ISEC2018 @CNRSenLR @INEE_CNRS #teamboulet @WWF @SMARTCnsvTools pic.twitter.com/FUiATuATNJ\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 5et 2018 ","date":1530748800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1530748800,"objectID":"0bf67949b78078c93ecb253793920e65","permalink":"https://oliviergimenez.github.io/blog/isec2018lm/","publishdate":"2018-07-05T00:00:00Z","relpermalink":"/blog/isec2018lm/","section":"blog","summary":"And nooooooow, ladies and gentlemen, @LMarescot presents her work on multispecies occupancy models to infer poaching #ISEC2018 @CNRSenLR @INEE_CNRS #teamboulet @WWF @SMARTCnsvTools pic.twitter.com/FUiATuATNJ\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 5et 2018 ","tags":["isec","statistical ecology"],"title":"International Statistical Ecology Conference 2018 (4/4) - Lucile Marescot","type":"blog"},{"authors":null,"categories":null,"content":"Aaaaaaaaaaaaaaaaaaaand now, ladies and gentlemen, @NSantostasi on stage to present her work on estimating prevalence of hybrids in 🇮🇹 wolves using #HMMs and capture-recapture data #teamboulet @CNRSenLR @INEE_CNRS @romolocaniglia @umontpellier #CEFE pic.twitter.com/vDXH9NDX2H\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 5 juillet 2018 ","date":1530662400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1530662400,"objectID":"49c1a21b9db564e6e3ebdb1c55f8b9c5","permalink":"https://oliviergimenez.github.io/blog/isec2018ns/","publishdate":"2018-07-04T00:00:00Z","relpermalink":"/blog/isec2018ns/","section":"blog","summary":"Aaaaaaaaaaaaaaaaaaaand now, ladies and gentlemen, @NSantostasi on stage to present her work on estimating prevalence of hybrids in 🇮🇹 wolves using #HMMs and capture-recapture data #teamboulet @CNRSenLR @INEE_CNRS @romolocaniglia @umontpellier #CEFE pic.","tags":["isec","statistical ecology"],"title":"International Statistical Ecology Conference 2018 (3/4) - Nina Santostasi","type":"blog"},{"authors":null,"categories":null,"content":"Are you ready for an awesome talk by @oaggimenez on regularization techniques for covariate selection in #caprecapt studies with Ian Renner @MoonbeamLevels ? #ISEC2018 @ISECStatEcol 😍💪🎉 pic.twitter.com/XD4IRPgoL8\n\u0026mdash; Julie Louvrier 🐱🐺 (@LouvrierJulie) 3 juillet 2018 ","date":1530576000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1530576000,"objectID":"7852a339665e4409e8269024d7022c9b","permalink":"https://oliviergimenez.github.io/blog/isec2018og/","publishdate":"2018-07-03T00:00:00Z","relpermalink":"/blog/isec2018og/","section":"blog","summary":"Are you ready for an awesome talk by @oaggimenez on regularization techniques for covariate selection in #caprecapt studies with Ian Renner @MoonbeamLevels ? #ISEC2018 @ISECStatEcol 😍💪🎉 pic.twitter.com/XD4IRPgoL8\n\u0026mdash; Julie Louvrier 🐱🐺 (@LouvrierJulie) 3 juillet 2018 ","tags":["isec","statistical ecology"],"title":"International Statistical Ecology Conference 2018 (2/4) - Olivier Gimenez","type":"blog"},{"authors":null,"categories":null,"content":"And now, ladies and gentlemen, the amazing @LouvrierJulie 👏 who tells us about mechanistic #SDMs to study 🐺 colonization in 🇫🇷 #ISEC2018 #teamboulet @ONCFS_officiel @CNRSenLR @INEE_CNRS @umontpellier pic.twitter.com/JUTayPLTSu\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 2 juillet 2018 This is the end... pic.twitter.com/2FNCT8yxib\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 2 juillet 2018 ","date":1530489600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1530489600,"objectID":"71a794473f6ae36080f22eb670aed86a","permalink":"https://oliviergimenez.github.io/blog/isec2018jl/","publishdate":"2018-07-02T00:00:00Z","relpermalink":"/blog/isec2018jl/","section":"blog","summary":"And now, ladies and gentlemen, the amazing @LouvrierJulie 👏 who tells us about mechanistic #SDMs to study 🐺 colonization in 🇫🇷 #ISEC2018 #teamboulet @ONCFS_officiel @CNRSenLR @INEE_CNRS @umontpellier pic.twitter.com/JUTayPLTSu\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 2 juillet 2018 This is the end.","tags":["isec","statistical ecology"],"title":"International Statistical Ecology Conference 2018 (1/4) - Julie Louvrier","type":"blog"},{"authors":null,"categories":null,"content":"Ma présentation aux Rencontres Humains \u0026amp; Cétacés en Méditerranée https://t.co/rkEhfC7TPu du 27 Juin 👧👦🤝🐋🐬 à #Marseille sur l\u0026#39;apport de la recherche académique https://t.co/MEZtmisGPJ @GIS_3M #teamboulet #CEFE @CNRSenLR @INEE_CNRS\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 2uin 2018 ","date":1529971200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1529971200,"objectID":"fc74929cce60214dedac1d9cfe477ed5","permalink":"https://oliviergimenez.github.io/blog/rhc/","publishdate":"2018-06-26T00:00:00Z","relpermalink":"/blog/rhc/","section":"blog","summary":"Ma présentation aux Rencontres Humains \u0026amp; Cétacés en Méditerranée https://t.co/rkEhfC7TPu du 27 Juin 👧👦🤝🐋🐬 à #Marseille sur l\u0026#39;apport de la recherche académique https://t.co/MEZtmisGPJ @GIS_3M #teamboulet #CEFE @CNRSenLR @INEE_CNRS\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 2uin 2018 ","tags":["conference","cetaceans","scicomm"],"title":"Rencontres Humains \u0026 Cétacés en Méditerranée","type":"blog"},{"authors":null,"categories":null,"content":"What future for 🐘 in Laos: adaptation or transformation. pic.twitter.com/RIbcPg4svz\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 19 juin 2018 ","date":1529366400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1529366400,"objectID":"91a66ba55b450c3fb1e125dac0151e63","permalink":"https://oliviergimenez.github.io/blog/gilles_maurer_phdefense/","publishdate":"2018-06-19T00:00:00Z","relpermalink":"/blog/gilles_maurer_phdefense/","section":"blog","summary":"What future for 🐘 in Laos: adaptation or transformation. pic.twitter.com/RIbcPg4svz\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 19 juin 2018 ","tags":["phd","gender balance"],"title":"Gilles Maurer's PhD defense on the interactions between captive \u0026 wild pops of 🐘 w/ an interdisciplinary approach","type":"blog"},{"authors":null,"categories":null,"content":"We\u0026#39;re lucky to have Anne Loison on June 18 at #CEFE who\u0026#39;ll tell us about the determinants of gender gaps in career trajectories #WomenInSTEM @INEE_CNRS @CNRSenLR @CNRS @FemmesCNRS #EgaliteFH @FemmesSciences @umontpellier @IsiteMUSE pic.twitter.com/u82uloqoIJ\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 19 mai 20\u003e Throught provoking seminar by Anne Loison @Univ_Savoie #LECA on \u0026quot;Professional and personal trajectories of 👩 🔬👨 🔬: ♂️♀️ differences among ecologists in 🇫🇷 and 🇳🇴\u0026quot; https://t.co/8adLzUGCPd w/ paper here https://t.co/6lqX0vxeiQ @CNRSenLR @INEE_CNRS #CEFE #teamboulet pic.twitter.com/wMaYPDnvll\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 18 juin 2018 I liked Anne\u0026#39;s idea of role models for men very much, and the quote from Henry David Thoreau: pic.twitter.com/Ng0ffaV72n\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 18 juin 2018 ","date":1529280000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1529280000,"objectID":"113f6cf9fd64501c1651570a7682058e","permalink":"https://oliviergimenez.github.io/blog/anne_loison/","publishdate":"2018-06-18T00:00:00Z","relpermalink":"/blog/anne_loison/","section":"blog","summary":"We\u0026#39;re lucky to have Anne Loison on June 18 at #CEFE who\u0026#39;ll tell us about the determinants of gender gaps in career trajectories #WomenInSTEM @INEE_CNRS @CNRSenLR @CNRS @FemmesCNRS #EgaliteFH @FemmesSciences @umontpellier @IsiteMUSE pic.twitter.com/u82uloqoIJ\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 19 mai 20\u003e ","tags":["conference","gender balance"],"title":"Determinants of gender gaps in science","type":"blog"},{"authors":null,"categories":null,"content":"Linking organismal functions, life history strategies and population performance @BritishEcolSoc cross-journal special feature by the flamboyant @DRobcito w/ @dylan_childs #CyrilleViolle \u0026amp; I https://t.co/qoNviB84ux w/ contribs by @HalCaswell @djm_MEEG @benjaminblonder \u0026amp; others 🤩 pic.twitter.com/fqd6FvBrPz\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 18 juin8 ","date":1529280000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1529280000,"objectID":"a87d4d94d7f22b4c0c27bdcc184c1d14","permalink":"https://oliviergimenez.github.io/blog/bibi3_paper/","publishdate":"2018-06-18T00:00:00Z","relpermalink":"/blog/bibi3_paper/","section":"blog","summary":"Linking organismal functions, life history strategies and population performance @BritishEcolSoc cross-journal special feature by the flamboyant @DRobcito w/ @dylan_childs #CyrilleViolle \u0026amp; I https://t.co/qoNviB84ux w/ contribs by @HalCaswell @djm_MEEG @benjaminblonder \u0026amp; others 🤩 pic.","tags":["traits","demography"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"Au siège du @CNRS pour présenter notre projet combinant #shs et #écologie pour la gestion des grands prédateurs 🐺🐻🐱 en 🇫🇷 \u0026quot;Osez l\u0026#39;#interdisciplinarité\u0026quot; 🤞😓 @CNRSenLR @INEE_CNRS pic.twitter.com/yxUFWj5cWB\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 11 juin 2018 Et la réponse vient de tomber, ça maaaaaaaaaaaarche! 💥🎉😎🥳 pic.twitter.com/g8lI7l1kJW\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 15 juin 2018 ","date":1528675200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1528675200,"objectID":"8854b59ecd2923740cd5a049149a31dc","permalink":"https://oliviergimenez.github.io/blog/ecosocar/","publishdate":"2018-06-11T00:00:00Z","relpermalink":"/blog/ecosocar/","section":"blog","summary":"Au siège du @CNRS pour présenter notre projet combinant #shs et #écologie pour la gestion des grands prédateurs 🐺🐻🐱 en 🇫🇷 \u0026quot;Osez l\u0026#39;#interdisciplinarité\u0026quot; 🤞😓 @CNRSenLR @INEE_CNRS pic.twitter.com/yxUFWj5cWB\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 11 juin 2018 Et la réponse vient de tomber, ça maaaaaaaaaaaarche!","tags":["interdisciplinarity","carnivores","social sciences","species distribution"],"title":"Socio-ecological approaches to better understand interactions between large carnivores and human activities","type":"blog"},{"authors":null,"categories":null,"content":"Building w/ stakeholders an interface to assess lynx viability in 🇫🇷 Sarah Bauduin gives a crash course on spatially-explicit individual-based models (aka PacMan) #rstats #shiny #leaflet https://t.co/v9e865t2ib #teamboulet pic.twitter.com/1oTtRQRG1j\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 6 juin 2018 ","date":1528243200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1528243200,"objectID":"cfa9080fa8eb201103633899dd79b9a9","permalink":"https://oliviergimenez.github.io/blog/ittecop_meeting2/","publishdate":"2018-06-06T00:00:00Z","relpermalink":"/blog/ittecop_meeting2/","section":"blog","summary":"Building w/ stakeholders an interface to assess lynx viability in 🇫🇷 Sarah Bauduin gives a crash course on spatially-explicit individual-based models (aka PacMan) #rstats #shiny #leaflet https://t.co/v9e865t2ib #teamboulet pic.twitter.com/1oTtRQRG1j\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 6 juin 2018 ","tags":["interdisciplinarity","lynx","wildife-vehicle collisions"],"title":"Workshop on mitigating lynx-vehicle collisions","type":"blog"},{"authors":null,"categories":null,"content":"Special feature on Evolutionary #Demography by Takenori Takada \u0026amp; @Richshefferson in #Population #Ecology @SpringerJP_LS - We wrote \u0026quot;Estimating individual fitness in the wild using capture-recapture data\u0026quot; https://t.co/5u7x1UvRr5 #rstats code ➡️ https://t.co/l8o7Igydn9 pic.twitter.com/IUp6egBPUA\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 3 juin8 ","date":1527984000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1527984000,"objectID":"45852860f1c3554d60513c31bc0d9c96","permalink":"https://oliviergimenez.github.io/blog/bibi_paper2/","publishdate":"2018-06-03T00:00:00Z","relpermalink":"/blog/bibi_paper2/","section":"blog","summary":"Special feature on Evolutionary #Demography by Takenori Takada \u0026amp; @Richshefferson in #Population #Ecology @SpringerJP_LS - We wrote \u0026quot;Estimating individual fitness in the wild using capture-recapture data\u0026quot; https://t.co/5u7x1UvRr5 #rstats code ➡️ https://t.","tags":["fitness","capture-recapture"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"Aux journées de la @Fondationfrance @univpaulvalery \u0026quot;La mer monte, quel littoral pour demain\u0026quot; à la belle ville d\u0026#39;#Agde pour présenter notre programme #Interact 🐬⛵🎣 avec @GIS_3M https://t.co/72YF9gxCeE pic.twitter.com/HbRf40p6PX\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 31 mai 2018ockquote\u003e ","date":1527724800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1527724800,"objectID":"73b42cdc8c0846a0d9bae6511903538d","permalink":"https://oliviergimenez.github.io/blog/interact_media2/","publishdate":"2018-05-31T00:00:00Z","relpermalink":"/blog/interact_media2/","section":"blog","summary":"Aux journées de la @Fondationfrance @univpaulvalery \u0026quot;La mer monte, quel littoral pour demain\u0026quot; à la belle ville d\u0026#39;#Agde pour présenter notre programme #Interact 🐬⛵🎣 avec @GIS_3M https://t.co/72YF9gxCeE pic.twitter.com/HbRf40p6PX \u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 31 mai 2018ockquote\u003e ","tags":["conference","interact","dolphin"],"title":"La mer monte, quel littoral pour demain","type":"blog"},{"authors":null,"categories":null,"content":"Thrilled to have @pardo_deborah for our #teamboulet 🤩👩 🔬 Freelance scientist and coordinator @PureOceanFund @HomewardBound16 #WomenInSTEM #womeninscience #worklife balance #Marseille #networking #scicomm #howtohavefivejobsatonce #outreach pic.twitter.com/t8OX4miuDB\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 29 mai 2018","date":1527552000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1527552000,"objectID":"09d0a5fab06148c007a012d4e602d76f","permalink":"https://oliviergimenez.github.io/blog/deb_pardo/","publishdate":"2018-05-29T00:00:00Z","relpermalink":"/blog/deb_pardo/","section":"blog","summary":"Thrilled to have @pardo_deborah for our #teamboulet 🤩👩 🔬 Freelance scientist and coordinator @PureOceanFund @HomewardBound16 #WomenInSTEM #womeninscience #worklife balance #Marseille #networking #scicomm #howtohavefivejobsatonce #outreach pic.twitter.com/t8OX4miuDB\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 29 mai 2018","tags":["conference","scicomm"],"title":"Team meeting on communicating science and outreach","type":"blog"},{"authors":null,"categories":null,"content":"Had a stimulating @AgenceRecherche meeting working on #multispecies #demographic models to study structure \u0026amp; dynamics of 🐺🐦🐟🌱 #communities Check out https://t.co/eNFOzrPgWR w/ @UMR_MECADEV @isemevol #CESCO #CEFE #ONCFS @MARBEC_MicMac #CEBC @CNRSenLR @INEE_CNRS #teamboulet pic.twitter.com/XcfVCBWq8L\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 24 mai18 ","date":1527120000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1527120000,"objectID":"4ecb4432c6a4dfa095a2f4b140b48815","permalink":"https://oliviergimenez.github.io/blog/anr_meeting2/","publishdate":"2018-05-24T00:00:00Z","relpermalink":"/blog/anr_meeting2/","section":"blog","summary":"Had a stimulating @AgenceRecherche meeting working on #multispecies #demographic models to study structure \u0026amp; dynamics of 🐺🐦🐟🌱 #communities Check out https://t.co/eNFOzrPgWR w/ @UMR_MECADEV @isemevol #CESCO #CEFE #ONCFS @MARBEC_MicMac #CEBC @CNRSenLR @INEE_CNRS #teamboulet pic.","tags":["ANR","democom","multispecies"],"title":"Project meeting on multispecies demography","type":"blog"},{"authors":null,"categories":null,"content":"We had a stimulating #teamboulet meeting this morning, w/ #decocom https://t.co/eNFOzrPgWR post-doc #LucileMarescot giving a crash course on how to make optimal decisions under #uncertainty using stochastic dynamic programming #SDP #decisionmaking @AgenceRecherche @CNRSenLR pic.twitter.com/XkdSYjFGrL\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 22 mai 2018 ","date":1526947200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1526947200,"objectID":"1761a1a738c6b3c4f35f464570dd8251","permalink":"https://oliviergimenez.github.io/blog/boulet_meeting5/","publishdate":"2018-05-22T00:00:00Z","relpermalink":"/blog/boulet_meeting5/","section":"blog","summary":"We had a stimulating #teamboulet meeting this morning, w/ #decocom https://t.co/eNFOzrPgWR post-doc #LucileMarescot giving a crash course on how to make optimal decisions under #uncertainty using stochastic dynamic programming #SDP #decisionmaking @AgenceRecherche @CNRSenLR pic.","tags":["team"],"title":"Team meeting on stochastic dynamic programming","type":"blog"},{"authors":null,"categories":null,"content":"Awesome #teamboulet weekly meeting last Friday with #ReginaBestbier \u0026amp; @AmandaVincent1 @projectseahorse @UBC on communicating science 👩‍🔬👨‍🔬 \u0026quot;1/ Who is your audience 2/ What is your message 3/ Keep it simple and easy\u0026quot; #scicomm #WomenInSTEM @CNRSenLR pic.twitter.com/vl1x8X4IjM\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 21 mai 2018 ","date":1526860800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1526860800,"objectID":"5e1ca405910a542a511075e7bc374e3c","permalink":"https://oliviergimenez.github.io/blog/boulet_meeting4/","publishdate":"2018-05-21T00:00:00Z","relpermalink":"/blog/boulet_meeting4/","section":"blog","summary":"Awesome #teamboulet weekly meeting last Friday with #ReginaBestbier \u0026amp; @AmandaVincent1 @projectseahorse @UBC on communicating science 👩‍🔬👨‍🔬 \u0026quot;1/ Who is your audience 2/ What is your message 3/ Keep it simple and easy\u0026quot; #scicomm #WomenInSTEM @CNRSenLR pic.","tags":["team"],"title":"Team meeting on communicating science","type":"blog"},{"authors":null,"categories":null,"content":"Very honored and pleased 🤩 to serve as an Associate Editor for the new @BritishEcolSoc journal \u0026#39;People and Nature\u0026#39; 🎉🍾 Looking forward to handling my first submission 👀 #SocialSciences #ecology #nature #humanities https://t.co/2jl8FDdNqc\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 11 mai 2018 ","date":1525996800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1525996800,"objectID":"ca0cffb190a8816b7605adeae872843b","permalink":"https://oliviergimenez.github.io/blog/people_nature/","publishdate":"2018-05-11T00:00:00Z","relpermalink":"/blog/people_nature/","section":"blog","summary":"Very honored and pleased 🤩 to serve as an Associate Editor for the new @BritishEcolSoc journal \u0026#39;People and Nature\u0026#39; 🎉🍾 Looking forward to handling my first submission 👀 #SocialSciences #ecology #nature #humanities https://t.","tags":["People and Nature"],"title":"Associate editor for People and Nature","type":"blog"},{"authors":null,"categories":null,"content":"Séminaire interfaces #linguistique et #écologie organisé par @Praxiling et #CEFE 14/05 @montpellier_ On y parlera de notre article récent sur le traitement médiatique du 🐺 dans la 🗞️ https://t.co/D8R2tzt3D0 @CNRSenLR @univpaulvalery @lemondefr @lemonde_science @Nice_Matin pic.twitter.com/k8RfAfOzQF\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 9 mai 2018 ","date":1525824000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1525824000,"objectID":"d19137c368941954a286a054aba618a4","permalink":"https://oliviergimenez.github.io/blog/sem_praxiling/","publishdate":"2018-05-09T00:00:00Z","relpermalink":"/blog/sem_praxiling/","section":"blog","summary":"Séminaire interfaces #linguistique et #écologie organisé par @Praxiling et #CEFE 14/05 @montpellier_ On y parlera de notre article récent sur le traitement médiatique du 🐺 dans la 🗞️ https://t.co/D8R2tzt3D0 @CNRSenLR @univpaulvalery @lemondefr @lemonde_science @Nice_Matin pic.","tags":["interdisciplinarity"],"title":"Séminaire interfaces linguistique et écologie","type":"blog"},{"authors":null,"categories":null,"content":"Science 🌉 policy: Inspiring meeting this morning w/ Sarah Fowler @saveourseas scientific adviser and @AmandaVincent1 \u0026quot;How scientists can influence policy? One person can make a difference\u0026quot; #teamboulet #CEFE @INEE_CNRS @CNRSenLR @umontpellier pic.twitter.com/S3J8jmz75d\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 7 mai 2018 ","date":1525651200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1525651200,"objectID":"cd0b67f64c48e4a00e9994f5ef452439","permalink":"https://oliviergimenez.github.io/blog/boulet_meeting3/","publishdate":"2018-05-07T00:00:00Z","relpermalink":"/blog/boulet_meeting3/","section":"blog","summary":"Science 🌉 policy: Inspiring meeting this morning w/ Sarah Fowler @saveourseas scientific adviser and @AmandaVincent1 \u0026quot;How scientists can influence policy? One person can make a difference\u0026quot; #teamboulet #CEFE @INEE_CNRS @CNRSenLR @umontpellier pic.","tags":["team"],"title":"Team meeting science and policy","type":"blog"},{"authors":null,"categories":null,"content":"\u0026quot;R2ucare: An R #rstats package to perform goodness‐of‐fit tests for capture‐recapture models\u0026quot; in @MethodsEcolEvol https://t.co/ieQohrn3jA 🤓😎 - paper entirely written w/ #rmarkdown thx @stevenvmiller https://t.co/9FapGe3oGV 💯 - replication files ➡️ https://t.co/h9F6sKbcpI pic.twitter.com/JyzQ33Wrl3\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 21 avril 2018 ","date":1524268800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1524268800,"objectID":"374ffda78aad73d73e4744f1928d4b55","permalink":"https://oliviergimenez.github.io/blog/bibi_paper/","publishdate":"2018-04-21T00:00:00Z","relpermalink":"/blog/bibi_paper/","section":"blog","summary":"\u0026quot;R2ucare: An R #rstats package to perform goodness‐of‐fit tests for capture‐recapture models\u0026quot; in @MethodsEcolEvol https://t.co/ieQohrn3jA 🤓😎 - paper entirely written w/ #rmarkdown thx @stevenvmiller https://t.co/9FapGe3oGV 💯 - replication files ➡️ https://t.","tags":["gof","capture-recapture"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"Ma thèse en 180 secondes (2018) - Des loups et des crottes 🐺💩 - Julie Louvrier https://t.co/DZMwANiEYD ou comment vulgariser son travail de #thèse sur la modélisation #statistique de la distribution des grands #carnivores #occupancy #SDM #MT180\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 16 avril 2018 ","date":1523836800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1523836800,"objectID":"cbfc67f8c887642766b09cec8c1df87b","permalink":"https://oliviergimenez.github.io/blog/julie_mt180/","publishdate":"2018-04-16T00:00:00Z","relpermalink":"/blog/julie_mt180/","section":"blog","summary":"Ma thèse en 180 secondes (2018) - Des loups et des crottes 🐺💩 - Julie Louvrier https://t.co/DZMwANiEYD ou comment vulgariser son travail de #thèse sur la modélisation #statistique de la distribution des grands #carnivores #occupancy #SDM #MT180","tags":["scicomm","wolf"],"title":"Julie dans MT180s","type":"blog"},{"authors":null,"categories":null,"content":"A career path from #academia to #datascience - inspiring #teamboulet meeting today w/ @StatnMap from @thinkR_fr. I\u0026#39;m sharing a few #careertips he gave us 💯 👇\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 3 avril 2018 ","date":1522800000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1522800000,"objectID":"cfbc006daf09f7bc07c7ceb8205be12b","permalink":"https://oliviergimenez.github.io/blog/team_boulet2/","publishdate":"2018-04-04T00:00:00Z","relpermalink":"/blog/team_boulet2/","section":"blog","summary":"A career path from #academia to #datascience - inspiring #teamboulet meeting today w/ @StatnMap from @thinkR_fr. I\u0026#39;m sharing a few #careertips he gave us 💯 👇\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 3 avril 2018 ","tags":["team","data science"],"title":"Team meeting: A career path from academia to data science","type":"blog"},{"authors":null,"categories":null,"content":"Awesome day of talks/discussions between the 🇫🇷 Game and Wildlife Agency #ONCFS and @INEE_CNRS to reinforce our links and improve the ways we work together @NirSeMa 🐹🐰🐻🐸🦊🐺🐦🐗🦌🦆 pic.twitter.com/DePhj2Qfnw\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 27 mars 2018 ","date":1522108800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1522108800,"objectID":"bbff32cac82a6c8686b681b5afc7ca2c","permalink":"https://oliviergimenez.github.io/blog/cnrs_oncfs_day/","publishdate":"2018-03-27T00:00:00Z","relpermalink":"/blog/cnrs_oncfs_day/","section":"blog","summary":"Awesome day of talks/discussions between the 🇫🇷 Game and Wildlife Agency #ONCFS and @INEE_CNRS to reinforce our links and improve the ways we work together @NirSeMa 🐹🐰🐻🐸🦊🐺🐦🐗🦌🦆 pic.twitter.com/DePhj2Qfnw\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 27 mars 2018 ","tags":["oncfs","wildlife management"],"title":"Journées d'échanges entre ONCFS et CNRS","type":"blog"},{"authors":null,"categories":null,"content":"Présentation de notre projet Interact https://t.co/72YF9gxCeE sur interactions dauphins-activités humaines dans le bulletin Pelagos 2018 https://t.co/rVHPhz1pnt @Fondationfrance @GIS_3M @Pelagos_ @INEE_CNRS @CNRSenLR pic.twitter.com/RYRlcxmYAq\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 20 mars 2018 ","date":1521590400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1521590400,"objectID":"ef058c2f3d0c0cefdc56c93413a286dc","permalink":"https://oliviergimenez.github.io/blog/interact_media/","publishdate":"2018-03-21T00:00:00Z","relpermalink":"/blog/interact_media/","section":"blog","summary":"Présentation de notre projet Interact https://t.co/72YF9gxCeE sur interactions dauphins-activités humaines dans le bulletin Pelagos 2018 https://t.co/rVHPhz1pnt @Fondationfrance @GIS_3M @Pelagos_ @INEE_CNRS @CNRSenLR pic.twitter.com/RYRlcxmYAq\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 20 mars 2018 ","tags":["interact","cetaceans","interdisciplinarity"],"title":"Interactions between dolphins and human activities","type":"blog"},{"authors":null,"categories":null,"content":"#multievent #capturerecapture workshop - Intro by #RogerPradel 😍📈 Check out the comments for a few relevant papers on the topic @INEE_CNRS @CNRSenLR pic.twitter.com/4nHrXBakgf\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 20 mars 2018 ","date":1521504000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1521504000,"objectID":"5818b15a0a7342ff5d03303b6dde2fcb","permalink":"https://oliviergimenez.github.io/blog/capturerecapture_workshop2018/","publishdate":"2018-03-20T00:00:00Z","relpermalink":"/blog/capturerecapture_workshop2018/","section":"blog","summary":"#multievent #capturerecapture workshop - Intro by #RogerPradel 😍📈 Check out the comments for a few relevant papers on the topic @INEE_CNRS @CNRSenLR pic.twitter.com/4nHrXBakgf\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 20 mars 2018 ","tags":["team","workshop","capture-recapture"],"title":"2018 capture-recapture workshop","type":"blog"},{"authors":null,"categories":null,"content":"Inspiring #teamboulet meeting on the role of scientists in public debate with special guest sociologist #LauraSeguin https://t.co/4DQyDYJ0vA #WomenInScience @FredBarraquand @AmandaVincent1 @ValentinLauret @SCubi25 pic.twitter.com/T4eY8hiN6H\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 17 mars 2018 ","date":1521244800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1521244800,"objectID":"33d7ef34f0ab046d02783c3a900ca057","permalink":"https://oliviergimenez.github.io/blog/boulet_meeting/","publishdate":"2018-03-17T00:00:00Z","relpermalink":"/blog/boulet_meeting/","section":"blog","summary":"Inspiring #teamboulet meeting on the role of scientists in public debate with special guest sociologist #LauraSeguin https://t.co/4DQyDYJ0vA #WomenInScience @FredBarraquand @AmandaVincent1 @ValentinLauret @SCubi25 pic.twitter.com/T4eY8hiN6H\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 17 mars 2018 ","tags":["team"],"title":"Team meeting on the role of scientists in public debate","type":"blog"},{"authors":null,"categories":null,"content":"Thrilled to welcome @AmandaVincent1 for a 6-month sabbatical in the #teamboulet @projectseahorse https://t.co/dcpXbRDJHB @INEE_CNRS @CNRSenLR pic.twitter.com/OyMbMKwO9K\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 17 mars 2018 ","date":1520380800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1520380800,"objectID":"8fc947f45f755245b2380f9f847016b0","permalink":"https://oliviergimenez.github.io/blog/amanda_sabbatical/","publishdate":"2018-03-07T00:00:00Z","relpermalink":"/blog/amanda_sabbatical/","section":"blog","summary":"Thrilled to welcome @AmandaVincent1 for a 6-month sabbatical in the #teamboulet @projectseahorse https://t.co/dcpXbRDJHB @INEE_CNRS @CNRSenLR pic.twitter.com/OyMbMKwO9K\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 17 mars 2018 ","tags":["interdisciplinarity","marine realm","conservation"],"title":"Amanda Vincent is on sabbatical with us","type":"blog"},{"authors":null,"categories":null,"content":"New paper 🎉🎉 \u0026quot;Social status mediates the fitness costs of infection with canine distemper virus in Serengeti spotted hyenas\u0026quot; https://t.co/LdWNAWFWum @BritishEcolSoc by L. Marescot and S. Benhaiem #teamboulet #multievent #hmm 😎😊 https://t.co/JKpN1scRTx pic.twitter.com/pD7yHGnhZ5\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 7 mars 2018 ","date":1520380800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1520380800,"objectID":"4d343cd3c78b6ce3da0959358cf820da","permalink":"https://oliviergimenez.github.io/blog/lulu_paper/","publishdate":"2018-03-07T00:00:00Z","relpermalink":"/blog/lulu_paper/","section":"blog","summary":"New paper 🎉🎉 \u0026quot;Social status mediates the fitness costs of infection with canine distemper virus in Serengeti spotted hyenas\u0026quot; https://t.co/LdWNAWFWum @BritishEcolSoc by L. Marescot and S. Benhaiem #teamboulet #multievent #hmm 😎😊 https://t.","tags":["interdisciplinarity","hyenas","capture-recapture"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"Journée prospective de notre laboratoire le #CEFE @INEE_CNRS @CNRSenLR @umontpellier 😍😉 pic.twitter.com/WwUDndaEgO\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 6 mars 2018 ","date":1520294400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1520294400,"objectID":"1ba968957849075938ddd9cb0fd36c58","permalink":"https://oliviergimenez.github.io/blog/prospective_cefe/","publishdate":"2018-03-06T00:00:00Z","relpermalink":"/blog/prospective_cefe/","section":"blog","summary":"Journée prospective de notre laboratoire le #CEFE @INEE_CNRS @CNRSenLR @umontpellier 😍😉 pic.twitter.com/WwUDndaEgO\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 6 mars 2018 ","tags":["CEFE"],"title":"Prospectives scientifiques du CEFE","type":"blog"},{"authors":null,"categories":null,"content":"Réunion de lancement avec les acteurs du territoire de notre projet sur \u0026quot;Eviter, réduire et compenser la mortalité du Lynx par collision avec les véhicules de transport\u0026quot; https://t.co/uKY5ylOli2 @_Cerema #oncfs @CROC_Estelle @INEE_CNRS #ittecop @ademe @FRBiodiv @Min_Ecologie 🤩💪 pic.twitter.com/3sY0rDIODQ\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 5 mars 2018 ","date":1520208000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1520208000,"objectID":"3b3f6c8cb1771f0100eddd62636ac33e","permalink":"https://oliviergimenez.github.io/blog/ittecop_meeting/","publishdate":"2018-03-05T00:00:00Z","relpermalink":"/blog/ittecop_meeting/","section":"blog","summary":"Réunion de lancement avec les acteurs du territoire de notre projet sur \u0026quot;Eviter, réduire et compenser la mortalité du Lynx par collision avec les véhicules de transport\u0026quot; https://t.co/uKY5ylOli2 @_Cerema #oncfs @CROC_Estelle @INEE_CNRS #ittecop @ademe @FRBiodiv @Min_Ecologie 🤩💪 pic.","tags":["interdisciplinarity","lynx","wildife-vehicle collisions"],"title":"Kick-off meeting of our project on mitigating lynx-vehicle collisions","type":"blog"},{"authors":null,"categories":null,"content":"New paper \u0026quot;Searching for meaning in marine mammal shared data\u0026quot; 🐬🖥️📈 #openaccess #reproducibility https://t.co/puZrXHACUW pic.twitter.com/7ULeAJLMfT\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 23 février 2018 ","date":1519344000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1519344000,"objectID":"2fd897cadb05e896ceb3c412d15ff043","permalink":"https://oliviergimenez.github.io/blog/gio_paper/","publishdate":"2018-02-23T00:00:00Z","relpermalink":"/blog/gio_paper/","section":"blog","summary":"New paper \u0026quot;Searching for meaning in marine mammal shared data\u0026quot; 🐬🖥️📈 #openaccess #reproducibility https://t.co/puZrXHACUW pic.twitter.com/7ULeAJLMfT\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 23 février 2018 ","tags":["interdisciplinarity","cetaceans","data sharing"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"New paper on \u0026quot;Content analysis of 🗞️ coverage of 🐺 recolonization in 🇫🇷 using structural topic modeling\u0026quot;; free PDF here: https://t.co/w4wHPfbSZ8 (there are production errors in the figures that should be corrected in a few days) pic.twitter.com/o8Qo9nY7zC\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 23 février 2018 ","date":1519344000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1519344000,"objectID":"3cd411348165eff5d0f07644f7d142bd","permalink":"https://oliviergimenez.github.io/blog/marie_paper/","publishdate":"2018-02-23T00:00:00Z","relpermalink":"/blog/marie_paper/","section":"blog","summary":"New paper on \u0026quot;Content analysis of 🗞️ coverage of 🐺 recolonization in 🇫🇷 using structural topic modeling\u0026quot;; free PDF here: https://t.co/w4wHPfbSZ8 (there are production errors in the figures that should be corrected in a few days) pic.","tags":["interdisciplinarity","wolf","content analysis","media"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"Philosophie, linguistique, géographie, psychologie, écologie : la question de la nature. #SHS dans un labo d écologie et d évolution! #femmesenscience #WomenInScience pic.twitter.com/mEQcur0EKv\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 16 février 2018 ","date":1518739200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1518739200,"objectID":"270346b48ad9793cdf4613acd0717ed0","permalink":"https://oliviergimenez.github.io/blog/biocon_meeting/","publishdate":"2018-02-16T00:00:00Z","relpermalink":"/blog/biocon_meeting/","section":"blog","summary":"Philosophie, linguistique, géographie, psychologie, écologie : la question de la nature. #SHS dans un labo d écologie et d évolution! #femmesenscience #WomenInScience pic.twitter.com/mEQcur0EKv\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 16 février 2018 ","tags":["interdisciplinarity","conservation","CEFE"],"title":"Annual meeting our Biodiversity and Conservation meeting","type":"blog"},{"authors":null,"categories":null,"content":" Recently, I have been using `OpenBUGS` for some analyses that `JAGS` cannot do. However, `JAGS` can be run in parallel through [the `jagsUI` package](https://github.com/kenkellner/jagsUI), which can save you some precious time. So the question is how to run several chains in parallel with `OpenBUGS`. Well, first you\u0026rsquo;ll need to install OpenBUGS (if you\u0026rsquo;re on a Mac, check out this short tutorial). Then, you\u0026rsquo;ll need to run OpenBUGS from R through the pacage R2OpenBUGS, which you can install via:\nif(!require(R2OpenBUGS)) install.packages(\u0026quot;R2OpenBUGS\u0026quot;) ## Loading required package: R2OpenBUGS Standard analysis Now let\u0026rsquo;s run the classical BUGS school example:\nLoad the OpenBUGS Package\nlibrary(R2OpenBUGS) Load the data\ndata(schools) Define the model, write it to a text file and have a look\nnummodel \u0026lt;- function(){ for (j in 1:J){ y[j] ~ dnorm (theta[j], tau.y[j]) theta[j] ~ dnorm (mu.theta, tau.theta) tau.y[j] \u0026lt;- pow(sigma.y[j], -2)} mu.theta ~ dnorm (0.0, 1.0E-6) tau.theta \u0026lt;- pow(sigma.theta, -2) sigma.theta ~ dunif (0, 1000) } write.model(nummodel, \u0026quot;nummodel.txt\u0026quot;) model.file1 = paste(getwd(),\u0026quot;nummodel.txt\u0026quot;, sep=\u0026quot;/\u0026quot;) file.show(\u0026quot;nummodel.txt\u0026quot;) Prepare the data for input into OpenBUGS\nJ \u0026lt;- nrow(schools) y \u0026lt;- schools$estimate sigma.y \u0026lt;- schools$sd data \u0026lt;- list (\u0026quot;J\u0026quot;, \u0026quot;y\u0026quot;, \u0026quot;sigma.y\u0026quot;) Initialization of variables\ninits \u0026lt;- function(){ list(theta = rnorm(J, 0, 100), mu.theta = rnorm(1, 0, 100), sigma.theta = runif(1, 0, 100))} Set the Wine working directory and the directory to OpenBUGS, and change the OpenBUGS.exe location as necessary:\nWINE=\u0026quot;/usr/local/Cellar/wine/2.0.4/bin/wine\u0026quot; WINEPATH=\u0026quot;/usr/local/Cellar/wine/2.0.4/bin/winepath\u0026quot; OpenBUGS.pgm=\u0026quot;/Applications/OpenBUGS323/OpenBUGS.exe\u0026quot; The are the parameters to save\nparameters = c(\u0026quot;theta\u0026quot;, \u0026quot;mu.theta\u0026quot;, \u0026quot;sigma.theta\u0026quot;) Run the model\nptm \u0026lt;- proc.time() schools.sim \u0026lt;- bugs(data, inits, model.file = model.file1,parameters=parameters,n.chains = 2, n.iter = 500000, n.burnin = 10000, OpenBUGS.pgm=OpenBUGS.pgm, WINE=WINE, WINEPATH=WINEPATH,useWINE=T) elapsed_time \u0026lt;- proc.time() - ptm elapsed_time ## user system elapsed ## 50.835 2.053 55.010 print(schools.sim) ## Inference for Bugs model at \u0026quot;/Users/oliviergimenez/Desktop/nummodel.txt\u0026quot;, ## Current: 2 chains, each with 5e+05 iterations (first 10000 discarded) ## Cumulative: n.sims = 980000 iterations saved ## mean sd 2.5% 25% 50% 75% 97.5% Rhat n.eff ## theta[1] 11.6 8.4 -1.9 6.1 10.5 15.8 32.0 1 370000 ## theta[2] 8.0 6.4 -4.8 4.0 8.0 12.0 20.9 1 67000 ## theta[3] 6.4 7.8 -11.3 2.2 6.8 11.2 20.9 1 55000 ## theta[4] 7.7 6.6 -5.7 3.7 7.8 11.8 20.9 1 70000 ## theta[5] 5.5 6.5 -8.8 1.6 5.9 9.8 17.1 1 26000 ## theta[6] 6.2 6.9 -8.9 2.3 6.6 10.7 18.9 1 23000 ## theta[7] 10.7 6.9 -1.4 6.0 10.1 14.7 26.2 1 480000 ## theta[8] 8.7 7.9 -6.8 4.0 8.4 13.0 25.7 1 76000 ## mu.theta 8.1 5.3 -2.0 4.7 8.1 11.4 18.5 1 30000 ## sigma.theta 6.6 5.7 0.2 2.5 5.2 9.1 20.9 1 12000 ## deviance 60.5 2.2 57.0 59.1 60.1 61.4 66.0 1 980000 ## ## For each parameter, n.eff is a crude measure of effective sample size, ## and Rhat is the potential scale reduction factor (at convergence, Rhat=1). ## ## DIC info (using the rule, pD = Dbar-Dhat) ## pD = 2.8 and DIC = 63.2 ## DIC is an estimate of expected predictive error (lower deviance is better). Parallel computations To run several chains in parallel, we\u0026rsquo;ll follow the steps described in this nice post.\n# loading packages library(snow) library(snowfall) # setting the number of CPUs to be 2 sfInit(parallel=TRUE, cpus=2) ## Warning in searchCommandline(parallel, cpus = cpus, type ## = type, socketHosts = socketHosts, : Unknown option on ## commandline: rmarkdown::render('/Users/oliviergimenez/Desktop/ ## run_openbugs_in_parallel.Rmd',~+~~+~encoding~+~ ## R Version: R version 3.4.3 (2017-11-30) ## snowfall 1.84-6.1 initialized (using snow 0.4-2): parallel execution on 2 CPUs. # and assigning the R2OpenBUGS library to each CPU sfLibrary(R2OpenBUGS) ## Library R2OpenBUGS loaded. ## Library R2OpenBUGS loaded in cluster. # create list of data J \u0026lt;- nrow(schools) y \u0026lt;- schools$estimate sigma.y \u0026lt;- schools$sd x.data \u0026lt;- list (J=J, y=y, sigma.y=sigma.y) # creating separate directory for each CPU process folder1 \u0026lt;- paste(getwd(), \u0026quot;/chain1\u0026quot;, sep=\u0026quot;\u0026quot;) folder2 \u0026lt;- paste(getwd(), \u0026quot;/chain2\u0026quot;, sep=\u0026quot;\u0026quot;) dir.create(folder1); dir.create(folder2); # sinking the model into a file in each directory for (folder in c(folder1, folder2)) { sink(paste(folder, \u0026quot;/nummodel.txt\u0026quot;, sep=\u0026quot;\u0026quot;)) cat(\u0026quot; model{ for (j in 1:J){ y[j] ~ dnorm (theta[j], tau.y[j]) theta[j] ~ dnorm (mu.theta, tau.theta) tau.y[j] \u0026lt;- pow(sigma.y[j], -2)} mu.theta ~ dnorm (0.0, 1.0E-6) tau.theta \u0026lt;- pow(sigma.theta, -2) sigma.theta ~ dunif (0, 1000) } \u0026quot;) sink() } # defining the function that will run MCMC on each CPU # Arguments: # chain - will be 1 or 2 # x.data - the data list # params - parameters to be monitored parallel.bugs \u0026lt;- function(chain, x.data, params) { # a. defining directory for each CPU sub.folder \u0026lt;- paste(getwd(),\u0026quot;/chain\u0026quot;, chain, sep=\u0026quot;\u0026quot;) # b. specifying the initial MCMC values inits \u0026lt;- function()list(theta = rnorm(x.data$J, 0, 100), mu.theta = rnorm(1, 0, 100), sigma.theta = runif(1, 0, 100)) # c. calling OpenBugs # (you may need to change the OpenBUGS.pgm directory) # je suis sous Mac, je fais tourner OpenBUGS via Wine bugs(data=x.data, inits=inits, parameters.to.save=params, n.iter = 500000, n.burnin = 10000, n.chains=1, model.file=\u0026quot;nummodel.txt\u0026quot;, debug=FALSE, codaPkg=TRUE, useWINE=TRUE, OpenBUGS.pgm = \u0026quot;/Applications/OpenBUGS323/OpenBUGS.exe\u0026quot;, working.directory = sub.folder, WINE=\u0026quot;/usr/local/Cellar/wine/2.0.4/bin/wine\u0026quot;, WINEPATH=\u0026quot;/usr/local/Cellar/wine/2.0.4/bin/winepath\u0026quot;) } # setting the parameters to be monitored params \u0026lt;- c(\u0026quot;theta\u0026quot;, \u0026quot;mu.theta\u0026quot;, \u0026quot;sigma.theta\u0026quot;) # calling the sfLapply function that will run # parallel.bugs on each of the 2 CPUs ptm \u0026lt;- proc.time() sfLapply(1:2, fun=parallel.bugs, x.data=x.data, params=params) ## [[1]] ## [1] \u0026quot;/Users/oliviergimenez/Desktop/chain1/CODAchain1.txt\u0026quot; ## ## [[2]] ## [1] \u0026quot;/Users/oliviergimenez/Desktop/chain2/CODAchain1.txt\u0026quot; elapsed_time = proc.time() - ptm elapsed_time ## user system elapsed ## 0.013 0.000 32.157 # locating position of each CODA chain chain1 \u0026lt;- paste(folder1, \u0026quot;/CODAchain1.txt\u0026quot;, sep=\u0026quot;\u0026quot;) chain2 \u0026lt;- paste(folder2, \u0026quot;/CODAchain1.txt\u0026quot;, sep=\u0026quot;\u0026quot;) # and, finally, getting the results res \u0026lt;- read.bugs(c(chain1, chain2)) ## Abstracting deviance ... 490000 valid values ## Abstracting mu.theta ... 490000 valid values ## Abstracting sigma.theta ... 490000 valid values ## Abstracting theta[1] ... 490000 valid values ## Abstracting theta[2] ... 490000 valid values ## Abstracting theta[3] ... 490000 valid values ## Abstracting theta[4] ... 490000 valid values ## Abstracting theta[5] ... 490000 valid values ## Abstracting theta[6] ... 490000 valid values ## Abstracting theta[7] ... 490000 valid values ## Abstracting theta[8] ... 490000 valid values ## Abstracting deviance ... 490000 valid values ## Abstracting mu.theta ... 490000 valid values ## Abstracting sigma.theta ... 490000 valid values ## Abstracting theta[1] ... 490000 valid values ## Abstracting theta[2] ... 490000 valid values ## Abstracting theta[3] ... 490000 valid values ## Abstracting theta[4] ... 490000 valid values ## Abstracting theta[5] ... 490000 valid values ## Abstracting theta[6] ... 490000 valid values ## Abstracting theta[7] ... 490000 valid values ## Abstracting theta[8] ... 490000 valid values summary(res) ## ## Iterations = 10001:5e+05 ## Thinning interval = 1 ## Number of chains = 2 ## Sample size per chain = 490000 ## ## 1. Empirical mean and standard deviation for each variable, ## plus standard error of the mean: ## ## Mean SD Naive SE Time-series SE ## deviance 60.453 2.221 0.002243 0.005737 ## mu.theta 8.109 5.261 0.005315 0.020596 ## sigma.theta 6.610 5.682 0.005740 0.027336 ## theta[1] 11.697 8.407 0.008493 0.027974 ## theta[2] 8.023 6.395 0.006460 0.019264 ## theta[3] 6.365 7.866 0.007946 0.022485 ## theta[4] 7.735 6.601 0.006668 0.019766 ## theta[5] 5.467 6.504 0.006570 0.022580 ## theta[6] 6.234 6.885 0.006955 0.021332 ## theta[7] 10.727 6.891 0.006961 0.023304 ## theta[8] 8.648 7.892 0.007972 0.021541 ## ## 2. Quantiles for each variable: ## ## 2.5% 25% 50% 75% 97.5% ## deviance 57.0200 59.120 60.040 61.430 65.99 ## mu.theta -2.0600 4.784 8.066 11.410 18.50 ## sigma.theta 0.2275 2.456 5.275 9.190 20.82 ## theta[1] -1.8850 6.195 10.560 15.880 32.04 ## theta[2] -4.8350 4.049 8.004 11.980 20.96 ## theta[3] -11.4800 2.194 6.871 11.170 20.90 ## theta[4] -5.7500 3.711 7.784 11.820 20.92 ## theta[5] -8.8490 1.603 5.938 9.834 17.14 ## theta[6] -8.8940 2.255 6.632 10.680 18.95 ## theta[7] -1.3450 6.125 10.140 14.680 26.27 ## theta[8] -6.8910 4.037 8.409 12.960 25.72 ","date":1515888000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1515888000,"objectID":"b5b5873064251d05885f45fb42d7bb69","permalink":"https://oliviergimenez.github.io/blog/run_openbugs_parallel/","publishdate":"2018-01-14T00:00:00Z","relpermalink":"/blog/run_openbugs_parallel/","section":"blog","summary":" Recently, I have been using `OpenBUGS` for some analyses that `JAGS` cannot do. However, `JAGS` can be run in parallel through [the `jagsUI` package](https://github.com/kenkellner/jagsUI), which can save you some precious time. So the question is how to run several chains in parallel with `OpenBUGS`.","tags":["bugs","R","rstats","OpenBUGS","parallel computation"],"title":"Running OpenBUGS in parallel","type":"blog"},{"authors":null,"categories":null,"content":" I had to use the good old `OpenBUGS` for some analyses that cannot be done in `JAGS`. Below are the steps to install `OpenBUGS` then to run it from your Mac either natively or from `R`. This tutorial is an adaptation of [this post](https://sites.google.com/site/mmeclimate/-bayesmet/openbugs-on-mac-os-x) and [that one](http://www.davideagle.org/r-2/bayesian-modeling-using-winbugs-and-openbugs/running-openbugs-on-mac-using-wine). If not done already, install Homebrew. This program will make the installation of any other programs on your Mac so easy!\nInstall Wine which will allow you to run any Windows programs (.exe) on your Mac. To do so, start by opening Terminal, then type in the command: brew install wine\nNext, download the Windows version of OpenBUGS here\nTo install OpenBUGS, still in Terminal, go to the directory where the file was downloaded and type (you might need to unzip the file you downloaded first): wine OpenBUGS323setup.exe\nOpenBUGS is now installed and ready to be used! You can run it by first going to the directory where OpenBUGS was installed. On my laptop, it can be achieved via the command: cd /Applications/OpenBUGS323\nThen, you just need to tye in the following command in the Terminal, and you should see an OpenBUGS windows poping up: wine OpenBUGS\nNow we would like to run OpenBUGS from R.\nInstall the package R2OpenBUGS by typing in the R console: if(!require(R2OpenBUGS)) install.packages(\u0026quot;R2OpenBUGS\u0026quot;) ## Loading required package: R2OpenBUGS Now let\u0026rsquo;s see whether everything works well by running the classical BUGS school example: Load the OpenBUGS Package\nlibrary(R2OpenBUGS) Load the data\ndata(schools) Define the model, write it to a text file and have a look\nnummodel \u0026lt;- function(){ for (j in 1:J){ y[j] ~ dnorm (theta[j], tau.y[j]) theta[j] ~ dnorm (mu.theta, tau.theta) tau.y[j] \u0026lt;- pow(sigma.y[j], -2)} mu.theta ~ dnorm (0.0, 1.0E-6) tau.theta \u0026lt;- pow(sigma.theta, -2) sigma.theta ~ dunif (0, 1000) } write.model(nummodel, \u0026quot;nummodel.txt\u0026quot;) model.file1 = paste(getwd(),\u0026quot;nummodel.txt\u0026quot;, sep=\u0026quot;/\u0026quot;) file.show(\u0026quot;nummodel.txt\u0026quot;) Prepare the data for input into OpenBUGS\nJ \u0026lt;- nrow(schools) y \u0026lt;- schools$estimate sigma.y \u0026lt;- schools$sd data \u0026lt;- list (\u0026quot;J\u0026quot;, \u0026quot;y\u0026quot;, \u0026quot;sigma.y\u0026quot;) Initialization of variables\ninits \u0026lt;- function(){ list(theta = rnorm(J, 0, 100), mu.theta = rnorm(1, 0, 100), sigma.theta = runif(1, 0, 100))} Set the Wine working directory and the directory to OpenBUGS, and change the OpenBUGS.exe location as necessary:\nWINE=\u0026quot;/usr/local/Cellar/wine/2.0.4/bin/wine\u0026quot; WINEPATH=\u0026quot;/usr/local/Cellar/wine/2.0.4/bin/winepath\u0026quot; OpenBUGS.pgm=\u0026quot;/Applications/OpenBUGS323/OpenBUGS.exe\u0026quot; The are the parameters to save\nparameters = c(\u0026quot;theta\u0026quot;, \u0026quot;mu.theta\u0026quot;, \u0026quot;sigma.theta\u0026quot;) Run the model\nschools.sim \u0026lt;- bugs(data, inits, model.file = model.file1,parameters=parameters,n.chains = 3, n.iter = 1000, OpenBUGS.pgm=OpenBUGS.pgm, WINE=WINE, WINEPATH=WINEPATH,useWINE=T) R will pause. You might get a weird message starting by err:ole, just ignore it. When the run is complete, a prompt will reappear, then just type the following command to get the result:\nprint(schools.sim) ## Inference for Bugs model at \u0026quot;/Users/oliviergimenez/Desktop/nummodel.txt\u0026quot;, ## Current: 3 chains, each with 1000 iterations (first 500 discarded) ## Cumulative: n.sims = 1500 iterations saved ## mean sd 2.5% 25% 50% 75% 97.5% Rhat n.eff ## theta[1] 12.2 7.9 -1.3 7.5 11.2 16.4 32.1 1.0 62 ## theta[2] 9.1 6.5 -4.0 5.1 9.4 13.2 21.4 1.0 150 ## theta[3] 7.8 7.7 -9.4 3.6 8.5 12.6 21.1 1.0 360 ## theta[4] 8.8 6.6 -4.5 4.5 9.2 13.3 20.4 1.0 110 ## theta[5] 6.8 6.9 -8.2 2.3 7.5 11.4 17.7 1.0 410 ## theta[6] 7.3 7.2 -8.6 2.7 8.2 11.8 18.9 1.0 190 ## theta[7] 11.5 6.4 -0.3 7.5 11.2 15.7 25.0 1.1 42 ## theta[8] 9.7 7.6 -4.7 5.1 9.6 14.4 25.1 1.0 130 ## mu.theta 9.2 5.2 -1.2 5.8 9.3 12.5 18.2 1.0 88 ## sigma.theta 5.9 5.6 0.2 1.7 4.4 8.5 20.2 1.1 51 ## deviance 60.7 2.2 57.2 59.2 60.1 61.9 65.6 1.0 120 ## ## For each parameter, n.eff is a crude measure of effective sample size, ## and Rhat is the potential scale reduction factor (at convergence, Rhat=1). ## ## DIC info (using the rule, pD = Dbar-Dhat) ## pD = 2.8 and DIC = 63.4 ## DIC is an estimate of expected predictive error (lower deviance is better). When run natively, WinBUGS and OpenBUGS have nice debugging capabilities; also, you can see what is going on, I mean the program reading the data, generating inits, and so on. To get the OpenBUGS window with a bunch of useful info, just add debug=T to the call of the bugs function, and re-run the model\nschools.sim \u0026lt;- bugs(data, inits, model.file = model.file1,parameters=parameters,n.chains = 3, n.iter = 1000, OpenBUGS.pgm=OpenBUGS.pgm, WINE=WINE, WINEPATH=WINEPATH,useWINE=T,debug=T) ## arguments 'show.output.on.console', 'minimized' and 'invisible' are for Windows only You will have to close the OpenBUGS window to get the prompt back.\n","date":1515801600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1515801600,"objectID":"dd8bf5c2cf56f37350447bdbe0526de5","permalink":"https://oliviergimenez.github.io/blog/run_openbugs_on_mac/","publishdate":"2018-01-13T00:00:00Z","relpermalink":"/blog/run_openbugs_on_mac/","section":"blog","summary":" I had to use the good old `OpenBUGS` for some analyses that cannot be done in `JAGS`. Below are the steps to install `OpenBUGS` then to run it from your Mac either natively or from `R`. This tutorial is an adaptation of [this post](https://sites.google.com/site/mmeclimate/-bayesmet/openbugs-on-mac-os-x) and [that one](http://www.davideagle.org/r-2/bayesian-modeling-using-winbugs-and-openbugs/running-openbugs-on-mac-using-wine).","tags":["bugs","R","rstats","OpenBUGS"],"title":"Run OpenBUGS on a Mac","type":"blog"},{"authors":null,"categories":null,"content":" Our project `Avoid, reduce and compensate mortality risks of the Eurasian Lynx by collision with transport vehicles` aims to build on previous works on lynx population viability, collision risks while crossing terrestrial transport infrastructures and mitigation strategies. Our team will develop an operational tool upon which the technical operators will be able to rely on in the decision-making process for land-use planning. More details [here](https://sites.google.com/view/erclynx/) (in French). This applied research project gathers a group of various actors (public and private; researchers, wildlife managers and carnivore experts; infrastructure managers and road/transport planning experts) with the following objectives:\nMaintaining functional connectivity between favorable lynx habitats and elaborating strategies for avoiding, reducing and compensating collision risks.\nBuilding a tool to reinforce and facilitation the implementation of land-use planning public policies.\nTo achieve these objectives, we will rely on:\nData on collisions (roads, railways) and external structural factors (ITT presence and their characteristics);\nA predictive statistical model to identify the zones with high probability of collision risks;\nA spatially-explicit individual-based population viability model (SEPVA) to predict extinction risks;\nA user-friendly computer program implemented in R. We will follow a companion modeling approach to ensure the smooth adoption and use of this program by the actors.\nThis project will fill a gap between the academic research, the stakes of land planning, the lynx conservation and the needs of ITT managers/operators. We will contribute to the implementation of national public policies (TVB, SRCE/SRADDET, \u0026ldquo;Avoid, Reduce, Compensate\u0026rdquo; doctrine) by testing our approach on pilot sites in the Vosges and Jura areas. This project will benefit from the sharing of experiences and skills from our colleagues in Switzerland and Germany.\n","date":1511827200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1511827200,"objectID":"4b1034113bd5b47e185f15c5e1455b62","permalink":"https://oliviergimenez.github.io/blog/erc_lynx/","publishdate":"2017-11-28T00:00:00Z","relpermalink":"/blog/erc_lynx/","section":"blog","summary":" Our project `Avoid, reduce and compensate mortality risks of the Eurasian Lynx by collision with transport vehicles` aims to build on previous works on lynx population viability, collision risks while crossing terrestrial transport infrastructures and mitigation strategies. Our team will develop an operational tool upon which the technical operators will be able to rely on in the decision-making process for land-use planning. More details [here](https://sites.google.com/view/erclynx/) (in French).","tags":["carnivores","management","lynx","software"],"title":"Terrestrial transport infrastructures, Lynx population viability and mitigation strategies","type":"blog"},{"authors":null,"categories":null,"content":" Here, I illustrate the possibility to use `JAGS` to simulate data with two examples that might be of interest to population ecologists: first a linear regression, second a Cormack-Jolly-Seber capture-recapture model to estimate animal survival (formulated as a state-space model). The code is available from [GitHub](https://github.com/oliviergimenez/simul_with_jags). Recently, I have been struggling with simulating data from complex hierarchical models. After several unsuccessful attempts in R, I remembered the good old times when I was using WinBUGS (more than 10 years already!) and the possibility to simulate data with it. I\u0026rsquo;m using JAGS now, and a quick search in Google with \u0026lsquo;simulating data with jags\u0026rsquo; led me to a complex example and a simple example.\nSimulating data with JAGS is convenient because you can use (almost) the same code for simulation and inference, and you can carry out simulation studies (bias, precision, interval coverage) in the same environment (namely JAGS).\nLinear regression example We first load the packages we need for this tutorial:\nlibrary(R2jags) library(runjags) library(mcmcplots) Then straight to the point, let\u0026rsquo;s generate data from a linear regression model. The trick is to use a data block, have the simplest model block you could think of and pass the parameters as if they were data. Note that it\u0026rsquo;d be possible to use only a model block, see comment here.\ntxtstring \u0026lt;- ' data{ # Likelihood: for (i in 1:N){ y[i] ~ dnorm(mu[i], tau) # tau is precision (1 / variance) mu[i] \u0026lt;- alpha + beta * x[i] } } model{ fake \u0026lt;- 0 } ' Here, alpha and beta are the intercept and slope, tau the precision or the inverse of the variance, y the response variable and x the explanatory variable.\nWe pick some values for the model parameters that we will use as data:\n# parameters for simulations N = 30 # nb of observations x \u0026lt;- 1:N # predictor alpha = 0.5 # intercept beta = 1 # slope sigma \u0026lt;- .1 # residual sd tau \u0026lt;- 1/(sigma*sigma) # precision # parameters are treated as data for the simulation step data\u0026lt;-list(N=N,x=x,alpha=alpha,beta=beta,tau=tau) Now call JAGS; note that we monitor the response variable instead of parameters as we would do when conducting standard inference:\n# run jags out \u0026lt;- run.jags(txtstring, data = data,monitor=c(\u0026quot;y\u0026quot;),sample=1, n.chains=1, summarise=FALSE) ## Compiling rjags model... ## Calling the simulation using the rjags method... ## Note: the model did not require adaptation ## Burning in the model for 4000 iterations... ## Running the model for 1 iterations... ## Simulation complete ## Finished running the simulation The output is a bit messy and needs to be formatted appropriately:\n# reformat the outputs Simulated \u0026lt;- coda::as.mcmc(out) Simulated ## Markov Chain Monte Carlo (MCMC) output: ## Start = 5001 ## End = 5001 ## Thinning interval = 1 ## y[1] y[2] y[3] y[4] y[5] y[6] y[7] y[8] ## 5001 1.288399 2.52408 3.61516 4.583587 5.600675 6.566052 7.593407 8.457497 ## y[9] y[10] y[11] y[12] y[13] y[14] y[15] y[16] ## 5001 9.70847 10.38035 11.5105 12.55048 13.49143 14.46356 15.45641 16.56148 ## y[17] y[18] y[19] y[20] y[21] y[22] y[23] ## 5001 17.50935 18.51501 19.66197 20.49477 21.57079 22.6199 23.48232 ## y[24] y[25] y[26] y[27] y[28] y[29] y[30] ## 5001 24.57923 25.47368 26.33674 27.46525 28.35525 29.60279 30.42952 dim(Simulated) ## [1] 1 30 dat = as.vector(Simulated) dat ## [1] 1.288399 2.524080 3.615160 4.583587 5.600675 6.566052 7.593407 ## [8] 8.457497 9.708470 10.380351 11.510500 12.550482 13.491435 14.463564 ## [15] 15.456410 16.561483 17.509350 18.515005 19.661969 20.494767 21.570790 ## [22] 22.619899 23.482317 24.579228 25.473676 26.336736 27.465251 28.355248 ## [29] 29.602791 30.429517 Now let\u0026rsquo;s fit the model we used to simulate to the data we just generated. I won\u0026rsquo;t go into the details and assume that the reader is familiar with JAGS and linear regression.\n# specify model in BUGS language model \u0026lt;- paste(\u0026quot;\tmodel { # Likelihood: for (i in 1:N){ y[i] ~ dnorm(mu[i], tau) # tau is precision (1 / variance) mu[i] \u0026lt;- alpha + beta * x[i] } # Priors: alpha ~ dnorm(0, 0.01) # intercept beta ~ dnorm(0, 0.01) # slope sigma ~ dunif(0, 100) # standard deviation tau \u0026lt;- 1 / (sigma * sigma) } \u0026quot;) writeLines(model,\u0026quot;lin_reg.jags\u0026quot;)\t# data jags.data \u0026lt;- list(y = dat, N = length(dat), x = x) # initial values inits \u0026lt;- function(){list(alpha = rnorm(1), beta = rnorm(1), sigma = runif(1,0,10))} # parameters monitored parameters \u0026lt;- c(\u0026quot;alpha\u0026quot;, \u0026quot;beta\u0026quot;, \u0026quot;sigma\u0026quot;) # MCMC settings ni \u0026lt;- 10000 nt \u0026lt;- 6 nb \u0026lt;- 5000 nc \u0026lt;- 2 # call JAGS from R res \u0026lt;- jags(jags.data, inits, parameters, \u0026quot;lin_reg.jags\u0026quot;, n.chains = nc, n.thin = nt, n.iter = ni, n.burnin = nb, working.directory = getwd()) ## module glm loaded ## Compiling model graph ## Resolving undeclared variables ## Allocating nodes ## Graph information: ## Observed stochastic nodes: 30 ## Unobserved stochastic nodes: 3 ## Total graph size: 130 ## ## Initializing model Let\u0026rsquo;s have a look to the results and compare with the parameters we used to simulate the data (see above):\n# summarize posteriors print(res, digits = 3) ## Inference for Bugs model at \u0026quot;lin_reg.jags\u0026quot;, fit using jags, ## 2 chains, each with 10000 iterations (first 5000 discarded), n.thin = 6 ## n.sims = 1668 iterations saved ## mu.vect sd.vect 2.5% 25% 50% 75% 97.5% Rhat ## alpha 0.544 0.038 0.469 0.518 0.545 0.570 0.617 1.000 ## beta 0.998 0.002 0.994 0.997 0.998 1.000 1.003 1.001 ## sigma 0.102 0.015 0.078 0.091 0.100 0.110 0.138 1.002 ## deviance -53.810 2.724 -56.867 -55.808 -54.516 -52.641 -46.676 1.001 ## n.eff ## alpha 1700 ## beta 1700 ## sigma 780 ## deviance 1700 ## ## For each parameter, n.eff is a crude measure of effective sample size, ## and Rhat is the potential scale reduction factor (at convergence, Rhat=1). ## ## DIC info (using the rule, pD = var(deviance)/2) ## pD = 3.7 and DIC = -50.1 ## DIC is an estimate of expected predictive error (lower deviance is better). Pretty close!\nCheck convergence:\n# trace plots traplot(res,c(\u0026quot;alpha\u0026quot;, \u0026quot;beta\u0026quot;, \u0026quot;sigma\u0026quot;)) Plot the posterior distribution of the regression parameters and residual standard deviation:\n# posterior distributions denplot(res,c(\u0026quot;alpha\u0026quot;, \u0026quot;beta\u0026quot;, \u0026quot;sigma\u0026quot;)) Capture-recapture example I now illustrate the use of JAGS to simulate data from a Cormack-Jolly-Seber model with constant survival and recapture probabilities. I assume that the reader is familiar with this model and its formulation as a state-space model.\nLet\u0026rsquo;s simulate!\ntxtstring \u0026lt;- ' data{ # Constant survival and recapture probabilities for (i in 1:nind){ for (t in f[i]:(n.occasions-1)){ phi[i,t] \u0026lt;- mean.phi p[i,t] \u0026lt;- mean.p } #t } #i # Likelihood for (i in 1:nind){ # Define latent state and obs at first capture z[i,f[i]] \u0026lt;- 1 mu2[i,1] \u0026lt;- 1 * z[i,f[i]] # detection is 1 at first capture (\u0026quot;conditional on first capture\u0026quot;) y[i,1] ~ dbern(mu2[i,1]) # then deal w/ subsequent occasions for (t in (f[i]+1):n.occasions){ # State process z[i,t] ~ dbern(mu1[i,t]) mu1[i,t] \u0026lt;- phi[i,t-1] * z[i,t-1] # Observation process y[i,t] ~ dbern(mu2[i,t]) mu2[i,t] \u0026lt;- p[i,t-1] * z[i,t] } #t } #i } model{ fake \u0026lt;- 0 } ' Let\u0026rsquo;s pick some values for parameters and store them in a data list:\n# parameter for simulations n.occasions = 10 # nb of occasions nind = 100 # nb of individuals mean.phi \u0026lt;- 0.8 # survival mean.p \u0026lt;- 0.6 # recapture f = rep(1,nind) # date of first capture data\u0026lt;-list(n.occasions = n.occasions, mean.phi = mean.phi, mean.p = mean.p, f = f, nind = nind) Now run JAGS:\nout \u0026lt;- run.jags(txtstring, data = data,monitor=c(\u0026quot;y\u0026quot;),sample=1, n.chains=1, summarise=FALSE) ## Compiling rjags model... ## Calling the simulation using the rjags method... ## Note: the model did not require adaptation ## Burning in the model for 4000 iterations... ## Running the model for 1 iterations... ## Simulation complete ## Finished running the simulation Format the output:\nSimulated \u0026lt;- coda::as.mcmc(out) dim(Simulated) ## [1] 1 1000 dat = matrix(Simulated,nrow=nind) head(dat) ## [,1] [,2] [,3] [,4] [,5] [,6] [,7] [,8] [,9] [,10] ## [1,] 1 1 0 0 0 0 0 0 0 0 ## [2,] 1 1 1 1 0 0 0 0 0 0 ## [3,] 1 0 0 0 0 0 0 0 0 0 ## [4,] 1 0 0 0 0 0 0 0 0 0 ## [5,] 1 0 0 0 0 0 0 0 0 0 ## [6,] 1 1 1 1 0 0 1 0 1 1 Here I monitored only the detections and non-detections, but it is also possible to get the simulated values for the states, i.e. whether an individual is alive or dead at each occasion. You just need to amend the call to JAGS with monitor=c(\u0026quot;y\u0026quot;,\u0026quot;x\u0026quot;) and to amend the output accordingly.\nNow we fit a Cormack-Jolly-Seber model to the data we\u0026rsquo;ve just simulated, assuming constant parameters:\nmodel \u0026lt;- paste(\u0026quot;\tmodel { # Priors and constraints for (i in 1:nind){ for (t in f[i]:(n.occasions-1)){ phi[i,t] \u0026lt;- mean.phi p[i,t] \u0026lt;- mean.p } #t } #i mean.phi ~ dunif(0, 1) # Prior for mean survival mean.p ~ dunif(0, 1) # Prior for mean recapture # Likelihood for (i in 1:nind){ # Define latent state at first capture z[i,f[i]] \u0026lt;- 1 for (t in (f[i]+1):n.occasions){ # State process z[i,t] ~ dbern(mu1[i,t]) mu1[i,t] \u0026lt;- phi[i,t-1] * z[i,t-1] # Observation process y[i,t] ~ dbern(mu2[i,t]) mu2[i,t] \u0026lt;- p[i,t-1] * z[i,t] } #t } #i } \u0026quot;) writeLines(model,\u0026quot;cjs.jags\u0026quot;)\tPrepare the data:\n# vector with occasion of marking get.first \u0026lt;- function(x) min(which(x!=0)) f \u0026lt;- apply(dat, 1, get.first) # data jags.data \u0026lt;- list(y = dat, f = f, nind = dim(dat)[1], n.occasions = dim(dat)[2]) # Initial values known.state.cjs \u0026lt;- function(ch){ state \u0026lt;- ch for (i in 1:dim(ch)[1]){ n1 \u0026lt;- min(which(ch[i,]==1)) n2 \u0026lt;- max(which(ch[i,]==1)) state[i,n1:n2] \u0026lt;- 1 state[i,n1] \u0026lt;- NA } state[state==0] \u0026lt;- NA return(state) } inits \u0026lt;- function(){list(mean.phi = runif(1, 0, 1), mean.p = runif(1, 0, 1), z = known.state.cjs(dat))} We\u0026rsquo;d like to carry out inference about survival and recapture probabilities:\nparameters \u0026lt;- c(\u0026quot;mean.phi\u0026quot;, \u0026quot;mean.p\u0026quot;) Standard MCMC settings:\nni \u0026lt;- 10000 nt \u0026lt;- 6 nb \u0026lt;- 5000 nc \u0026lt;- 2 Ready to run JAGS!\n# Call JAGS from R (BRT 1 min) cjs \u0026lt;- jags(jags.data, inits, parameters, \u0026quot;cjs.jags\u0026quot;, n.chains = nc, n.thin = nt, n.iter = ni, n.burnin = nb, working.directory = getwd()) ## Compiling model graph ## Resolving undeclared variables ## Allocating nodes ## Graph information: ## Observed stochastic nodes: 900 ## Unobserved stochastic nodes: 902 ## Total graph size: 3707 ## ## Initializing model Summarize posteriors and compare to the values we used to simulate the data:\nprint(cjs, digits = 3) ## Inference for Bugs model at \u0026quot;cjs.jags\u0026quot;, fit using jags, ## 2 chains, each with 10000 iterations (first 5000 discarded), n.thin = 6 ## n.sims = 1668 iterations saved ## mu.vect sd.vect 2.5% 25% 50% 75% 97.5% Rhat ## mean.p 0.596 0.033 0.531 0.574 0.597 0.618 0.660 1.000 ## mean.phi 0.784 0.021 0.742 0.770 0.785 0.799 0.824 1.001 ## deviance 440.611 18.374 408.121 427.569 438.662 452.512 479.608 1.001 ## n.eff ## mean.p 1700 ## mean.phi 1700 ## deviance 1700 ## ## For each parameter, n.eff is a crude measure of effective sample size, ## and Rhat is the potential scale reduction factor (at convergence, Rhat=1). ## ## DIC info (using the rule, pD = var(deviance)/2) ## pD = 168.9 and DIC = 609.5 ## DIC is an estimate of expected predictive error (lower deviance is better). Again pretty close!\nTrace plots\ntraplot(cjs,c(\u0026quot;mean.phi\u0026quot;, \u0026quot;mean.p\u0026quot;)) Posterior distribution plots:\ndenplot(cjs,c(\u0026quot;mean.phi\u0026quot;, \u0026quot;mean.p\u0026quot;)) ","date":1511395200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1511395200,"objectID":"d4d9b9c66c44b9abcde82ffc038aa760","permalink":"https://oliviergimenez.github.io/blog/sim_with_jags/","publishdate":"2017-11-23T00:00:00Z","relpermalink":"/blog/sim_with_jags/","section":"blog","summary":" Here, I illustrate the possibility to use `JAGS` to simulate data with two examples that might be of interest to population ecologists: first a linear regression, second a Cormack-Jolly-Seber capture-recapture model to estimate animal survival (formulated as a state-space model). The code is available from [GitHub](https://github.com/oliviergimenez/simul_with_jags).","tags":["rstats","R","JAGS","simulations"],"title":"Simulating data with JAGS","type":"blog"},{"authors":null,"categories":null,"content":" Following my recent attempt to [fit a HMM model to capture-recapture data with TMB](https://oliviergimenez.github.io/post/multievent_in_tmb/) and the rather estonishing outcome (the code was \u003e 300 time faster than the equivalent R code!), I was curious to add TMB to the [list of options I tried to fit dynamic occupancy models](https://oliviergimenez.github.io/post/occupancy_in_admb/). Well, the least I can say is that TMB is fast, damn fast! The reasons for trying TMB were the same as before: TMB is said to be fast, allows for parallel computations, works with R, accomodates spatial stuff, allows easy implementation of random effects).\nI found materials on the internet to teach myself TMB, at least what I needed to implement a simple HMM model. See here for a linear regression and a Gompertz state space model examples, here for the same linear regression example on Youtube (that\u0026rsquo;s awesome!) and many other examples here.\nThe R code is available on my GitHub here. TMB was\u0026hellip; wait for it\u0026hellip; \u0026gt; 300 times faster than ADMB, \u0026gt; 140 times than Unmarked and \u0026gt; 6000 times faster than Jags (although the comparison with the latter is a bit unfair I suppose). The results are available here.\nI\u0026rsquo;m new to TMB, but I\u0026rsquo;m gonna definitely dig into it. Congrats to the developers! Check out the TMB website as well as the paper that comes with it.\n","date":1503576000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1503576000,"objectID":"2fab12daada98d56b5bcac98e364aa1b","permalink":"https://oliviergimenez.github.io/blog/occupancy_in_tmb/","publishdate":"2017-08-24T12:00:00Z","relpermalink":"/blog/occupancy_in_tmb/","section":"blog","summary":" Following my recent attempt to [fit a HMM model to capture-recapture data with TMB](https://oliviergimenez.github.io/post/multievent_in_tmb/) and the rather estonishing outcome (the code was \u003e 300 time faster than the equivalent R code!), I was curious to add TMB to the [list of options I tried to fit dynamic occupancy models](https://oliviergimenez.github.io/post/occupancy_in_admb/). Well, the least I can say is that TMB is fast, damn fast!","tags":["occupancy","R","hmm","TMB"],"title":"Fitting dynamic occupancy models with TMB","type":"blog"},{"authors":null,"categories":null,"content":" Following my attempts to fit a HMM model to [capture-recapture data with Rcpp](http://localhost:1313/post/multievent_in_rcpp/) and to [occupancy data with ADMB](http://localhost:1313/post/occupancy_in_admb/), a few colleagues suggested TMB as a potential alternative for several reasons (fast, allows for parallel computations, works with R, accomodates spatial stuff, easy implementation of random effects, and probably other reasons that I don't know). I found materials on the internet to teach myself TMB, at least what I needed to implement a simple HMM model. See here for a linear regression and a Gompertz state space model examples, here for the same linear regression example on Youtube (that\u0026rsquo;s awesome!) and many other examples here. However, I got stuck and posted my desperate request for help on the TMB forum. Guess what, I got an answer less than a few hours after - thank you Mollie Brooks!\nThe R code is available on my GitHub here. Minimizing the deviance coded with TMB was\u0026hellip; wait for it\u0026hellip; \u0026gt; 300 times faster than using the deviance coded in standard R.\nHope this is useful.\n","date":1503316800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1503316800,"objectID":"5224317b772b568ef4261a60c1e4fd41","permalink":"https://oliviergimenez.github.io/blog/multievent_in_tmb/","publishdate":"2017-08-21T12:00:00Z","relpermalink":"/blog/multievent_in_tmb/","section":"blog","summary":" Following my attempts to fit a HMM model to [capture-recapture data with Rcpp](http://localhost:1313/post/multievent_in_rcpp/) and to [occupancy data with ADMB](http://localhost:1313/post/occupancy_in_admb/), a few colleagues suggested TMB as a potential alternative for several reasons (fast, allows for parallel computations, works with R, accomodates spatial stuff, easy implementation of random effects, and probably other reasons that I don't know).","tags":["capture-recapture","multievent","R","hmm","TMB"],"title":"Fitting HMM/multievent capture-recapture models with TMB","type":"blog"},{"authors":null,"categories":null,"content":" Over the years (age privilege), I sat on several hiring committees for French universities and research institutes. I also have students who are now in the position to apply for such positions. It had me thinking about what works and what doesn't work in job applications and interviews. Here are my 2 cents suggestions. You can find plenty of more clever and relevant recommendations on the internet (see here, here or here for example). Also, be aware that these recommendations are specific to the French system, with little relevance for other countries.\nYour application\nRead and stick to the guidelines (not that obvious sometimes). The French academic system has some specific requirements to apply, in particular for faculty positions. Try to connect your CV with the job profile.\nDo not hesitate to include figures and boxes in the document. Structure does not hurt, well articulated sections are welcome. Be crystal clear on your current position. Synthetic tables summarizing the courses you taught, the talks you gave, the papers you published will help the referees to grab what is important in a minute (they obviously have more than your application to evaluate). Distinguish papers from oral presentations.\nSubmitted papers do not \u0026lsquo;count\u0026rsquo;. Use preprint servers such as arXiv or bioRxiv so that your referees get a chance to read what you\u0026rsquo;ve been working on lately (and judge the quality by themselves).\nProject yourself in the team/lab where the researcher will be hosted, and in the job itself. Feel free to get in touch with the host team/lab. Try to meet people beforehand. Visit websites. Get to know your hopefully-soon-to-be new lab. Show you\u0026rsquo;re interested in working with your hopefully-soon-to-be new colleagues. This might also help you find out about the \u0026lsquo;dismal job that everyone has been avoiding for years\u0026rsquo;.\nFor lecturer positions, even though it is not required, write a short project section including research objectives and do not forget your teaching objectives!\nHave your applications read by one or two colleagues, if possible someone who has experience in being on the kind of hiring committee you will have to face.\nYour interview\nBe on site the day before ( strikes in France happen).\nAppearances do matter, whether you like it or not, therefore pick something neutral and comfortable (my opinion).\nDo not use slides with too much text or complex figures (no tables please!), and point at your slides so that your visual support is part of your speech. Don\u0026rsquo;t look at the screen or the walls, try to look at the members of the jury. Occupy the space, don\u0026rsquo;t always stay at the same spot, but do not move too much (I\u0026rsquo;m useless at that).\nDo not rush when you talk, and give short answers so that every member of the committee gets a chance to ask you a question. Be honest: if you don\u0026rsquo;t know something, or if you don\u0026rsquo;t understand a question (which most likely means that members of the hiring committee do not understand this question either), say it.\nLast, but not least: rehearse, rehearse and rehearse again, there will be a moment where you do no longer sound like you know your text by heart, but rather, like saying it has become natural.\nFrom my experience, the round of questions is more determinant than the talk itself, but we have very rare opportunities to train on how to answer questions. Try to have sessions of questions with your colleagues to get better at it; it comes with experience, no secret.\nFor short talks in general, my advice is to write your text, at least for the first slides so that you\u0026rsquo;re confident you won\u0026rsquo;t get stuck when you start giving your speech. Make a video of yourself rehearsing and watch it; this is painful (I know by experience), but helpful.\nShow that you\u0026rsquo;re excited by the job offer, and that you like teaching and doing research. At the end of the day, I (sadly) vividly remember only a few candidates, most often those who demonstrated interest in the job and pleasure in doing what they do. I wasn\u0026rsquo;t good at inspiring that myself, but one thing that helped me was to think hard why I was applying, and to convince myself that I wanted the job so badly that I had to convince the committee that they should offer the job to me.\nLast but not least, do not do the job of the committee for the committee. In other words, do not censor yourself. Apply, and let the referees do their job.\nGood luck!\n","date":1502712000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1502712000,"objectID":"67174cc607328aa6d1a8b49954e4f365","permalink":"https://oliviergimenez.github.io/blog/interview_inra/","publishdate":"2017-08-14T12:00:00Z","relpermalink":"/blog/interview_inra/","section":"blog","summary":" Over the years (age privilege), I sat on several hiring committees for French universities and research institutes. I also have students who are now in the position to apply for such positions. It had me thinking about what works and what doesn't work in job applications and interviews. Here are my 2 cents suggestions.","tags":["interview","tenure","academia","university","application"],"title":"Applying and interviewing for an academic job","type":"blog"},{"authors":null,"categories":null,"content":"","date":1502496060,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1502496060,"objectID":"66f56322f3bd540dc2c42b4bbb3ff06a","permalink":"https://oliviergimenez.github.io/my-project/external-project/interact/","publishdate":"2017-08-12T00:01:00Z","relpermalink":"/my-project/external-project/interact/","section":"my-project","summary":"Combining ecology and social sciences to study interactions between human activities and bottlenose dolphins in the French Mediterranean sea.","tags":["sociology","cetaceans","management","interdisciplinary"],"title":"Human-dolphin interactions","type":"my-project"},{"authors":null,"categories":null,"content":"","date":1502496000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1502496000,"objectID":"2499b7cc7137f7de5df3181891fd4f9b","permalink":"https://oliviergimenez.github.io/my-project/external-project/democom/","publishdate":"2017-08-12T00:00:00Z","relpermalink":"/my-project/external-project/democom/","section":"my-project","summary":"Developing multispecies demography to assess the effects of climate and management on community dynamics.","tags":["demography","ANR","management","climate"],"title":"Multispecies demography","type":"my-project"},{"authors":null,"categories":null,"content":" Following my previous post on [using ADMB to fit hidden Markov models](https://oliviergimenez.github.io/post/occupancy_in_admb/), I took some time to learn how to use Rcpp ([Eddelbuettel \u0026 Francois 2011](https://www.jstatsoft.org/article/view/v040i08); [Eddelbuettel 2013](http://www.springer.com/us/book/9781461468677)), a package that gives friendly access to the power of C++ and increase the speed of your R programs. Kudos to Dirk Eddelbuettel, Romain Francois and their colleagues, Rcpp is awesome! I started with the excellent Rcpp chapter in the Advanced R book by Hadley Wickham which I complemented with the various vignettes that come with the package. As always, I googled the problems I had and often ended up finding the solution on stackoverflow. The rcpp-devel discussion list is the place where questions should be asked about Rcpp.\nMy objective was to implement the likelihood of a relatively simple multievent capture-recapture model ( Pradel 2005) with Rcpp. I recycled some R code I had and a dataset on shearwaters I used in a paper ( Gimenez et al. 2012).\nThe R code is available on my GitHub here. To run it, you just need to type Rcpp::sourceCpp(\u0026lsquo;multi event.cpp\u0026rsquo;) in the console. I\u0026rsquo;m convinced that the code can be improved, but this simple exercise showed that minimizing the deviance coded with Rcpp and calculating the Hessian was 10 times faster than using the deviance coded in standard R.\nNext steps will be to go for RcppArmadillo for matrix computations and RcppNumerical for optimisation (and numerical integration for random effects).\nHope this is useful.\nUpdate: Following an advice from Romain Francois and Dirk Eddelbuettel (the Rcpp gurus), I have switched to RcppArmadillo to rely on the code developed by professionals and decades of testing. Now the RcppArmadillo code is 50 times faster than basic R! The code is available on my GitHub.\n","date":1502452800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1502452800,"objectID":"f5fa1210c972854031fd4113fe0c2471","permalink":"https://oliviergimenez.github.io/blog/multievent_in_rcpp/","publishdate":"2017-08-11T12:00:00Z","relpermalink":"/blog/multievent_in_rcpp/","section":"blog","summary":" Following my previous post on [using ADMB to fit hidden Markov models](https://oliviergimenez.github.io/post/occupancy_in_admb/), I took some time to learn how to use Rcpp ([Eddelbuettel \u0026 Francois 2011](https://www.jstatsoft.org/article/view/v040i08); [Eddelbuettel 2013](http://www.springer.com/us/book/9781461468677)), a package that gives friendly access to the power of C++ and increase the speed of your R programs. Kudos to Dirk Eddelbuettel, Romain Francois and their colleagues, Rcpp is awesome!","tags":["capture-recapture","multievent","R","hmm"],"title":"Fitting multievent capture-recapture models with Rcpp","type":"blog"},{"authors":null,"categories":null,"content":" Some time ago, a student of mine got stuck when fitting dynamic occupancy models to real data in Jags because of the computational burden. We had a dataset with several thousands sites, more than 20 seasons and 4 surveys per season (yeah!).\nWe thought of using Unmarked instead (the likelihood is written in C++ and used through Rcpp), but dynamic models with false positives and/or random effects are not (yet?) implemented, and we were interested in considering both in our analysis. Some years ago, I had the opportunity to learn ADMB in a NCEAS meeting (thanks Hans Skaug!), I thought I would give it a try. ADMB allows you to write down any likelihood functions yourself and to incorporate random effects in an efficient way. It\u0026rsquo;s known to be fast for reasons I won\u0026rsquo;t go into here. Last but not least, ADMB can be run from R like JAGS and Unmarked (thanks Ben Bolker!).\nHere we go. I first simulate some data, then fit a dynamic model using ADMB, JAGS and Unmarked and finally perform a quick benchmarking. I\u0026rsquo;m going for a standard dynamic model, because the aims are i) to verify that JAGS is slower than Unmarked, ii) that ADMB is closer to Unmarked than JAGS in terms of time computation. If ii) is verified, then it will be worth the effort coding everything in ADMB.\nThe results are available on RPub here. The code is available on GitHub here.\nHope this is useful.\n","date":1502020800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1502020800,"objectID":"ea126ad1d912f19f4c3d55c5b7bb5e1b","permalink":"https://oliviergimenez.github.io/blog/occupancy_in_admb/","publishdate":"2017-08-06T12:00:00Z","relpermalink":"/blog/occupancy_in_admb/","section":"blog","summary":" Some time ago, a student of mine got stuck when fitting dynamic occupancy models to real data in Jags because of the computational burden.","tags":["jags","occupancy","R","hmm"],"title":"Fitting occupancy models in ADMB","type":"blog"},{"authors":null,"categories":null,"content":" I just submitted to Oikos a paper entitled \"Individual heterogeneity and capture-recapture models: what, why and how?\". A pre-print is available [here](http://biorxiv.org/content/early/2017/03/26/120683). I also provide R codes on GitHub [there](http://bit.ly/2nqWzoZ). This is joint work with Emmanuelle Cam and Jean-Michel Gaillard. This work was greatly inspired by my HDR I defended back in 2010 (!) and the ANR project I got over the period 2008-2011 (re-!).\n","date":1490616000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1490616000,"objectID":"bc19b0c3d2bfdb2fd8be08b28f610c23","permalink":"https://oliviergimenez.github.io/blog/paper_idh/","publishdate":"2017-03-27T12:00:00Z","relpermalink":"/blog/paper_idh/","section":"blog","summary":" I just submitted to Oikos a paper entitled \"Individual heterogeneity and capture-recapture models: what, why and how?\". A pre-print is available [here](http://biorxiv.org/content/early/2017/03/26/120683). I also provide R codes on GitHub [there](http://bit.ly/2nqWzoZ).","tags":["paper","individual heterogeneity","capture-recapture"],"title":"Review on individual heterogeneity in capture-recapture models","type":"blog"},{"authors":null,"categories":null,"content":" Our report (in French) on foresighting wolf population dynamic in France is available from [here](https://t.co/a6ibQoF7Nr). My contribution was mostly based on the work of my present or former students, including L. Marescot, J. Louvrier and S. Cubaynes. Difficult exercise, partly because you constantly think of what people will retain from your writings.\n","date":1490443200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1490443200,"objectID":"937ba394171c2cbb64a75b5a0b66217b","permalink":"https://oliviergimenez.github.io/blog/wolf_expertise/","publishdate":"2017-03-25T12:00:00Z","relpermalink":"/blog/wolf_expertise/","section":"blog","summary":" Our report (in French) on foresighting wolf population dynamic in France is available from [here](https://t.co/a6ibQoF7Nr). My contribution was mostly based on the work of my present or former students, including L. Marescot, J. Louvrier and S. Cubaynes.","tags":["expertise","wolf","population dynamics"],"title":"Expertise on wolf population dynamic in France","type":"blog"},{"authors":null,"categories":null,"content":" We held our annual 5-day [capture-recapture workshop](https://multievent.sciencesconf.org/) in Montpellier. As always, the attendees had challenging questions and stimulating datasets to analyse! Looking forward to next year. ","date":1490011200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1490011200,"objectID":"e757f1fb972fe1bc3fba8985e8ebfb01","permalink":"https://oliviergimenez.github.io/blog/cr_workshop_2017/","publishdate":"2017-03-20T12:00:00Z","relpermalink":"/blog/cr_workshop_2017/","section":"blog","summary":" We held our annual 5-day [capture-recapture workshop](https://multievent.sciencesconf.org/) in Montpellier. As always, the attendees had challenging questions and stimulating datasets to analyse! Looking forward to next year. ","tags":["workshop","capture-recapture"],"title":"Capture-recapture workshop in Montpellier","type":"blog"},{"authors":null,"categories":null,"content":" Isabelle Arpin, a sociologist from Grenoble, organised a 2-day seminar on interdisciplinarity. The idea was for the participants to share their experience in practicing interdisciplinarity. Coralie Mounet, Nicolas Lescureux and I gave a talk on our collaboration on the attitude towards brown bears.\n","date":1489406400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1489406400,"objectID":"2bfa767f513559115bf7c18d08a7db8b","permalink":"https://oliviergimenez.github.io/blog/seminar_grenoble/","publishdate":"2017-03-13T12:00:00Z","relpermalink":"/blog/seminar_grenoble/","section":"blog","summary":" Isabelle Arpin, a sociologist from Grenoble, organised a 2-day seminar on interdisciplinarity. The idea was for the participants to share their experience in practicing interdisciplinarity.","tags":["conference","interdisciplinarity","social sciences"],"title":"Seminar on interdisciplinarity in practice","type":"blog"},{"authors":null,"categories":null,"content":" I spent 3 days in Berlin at the [Leibniz Institute for Zoo and Wildlife Research](http://www.leibniz-izw.de/welcome.html) catching up with friends and colleagues (L. Marescot, S. Benhaim, S. Kramer-Schadt, A. Courtiol, H. Hofer). We talked about spotted hyenas and monitoring of large carnivores. I also gave a workshop on inferring occupancy using HMM. I even got in touch with an old acquaintance - Steve Beissinger - who happens to be in Berlin for the year. Steve will come and visit us in Montpellier for a few days in May.\n","date":1488974400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1488974400,"objectID":"ea32ac8f546fe118547c2b9b960dcdea","permalink":"https://oliviergimenez.github.io/blog/visit_berlin_2017/","publishdate":"2017-03-08T12:00:00Z","relpermalink":"/blog/visit_berlin_2017/","section":"blog","summary":" I spent 3 days in Berlin at the [Leibniz Institute for Zoo and Wildlife Research](http://www.leibniz-izw.de/welcome.html) catching up with friends and colleagues (L. Marescot, S. Benhaim, S. Kramer-Schadt, A. Courtiol, H. Hofer).","tags":["workshop","Berlin","trip"],"title":"Visit to the IZW Leibniz Institute, Berlin","type":"blog"},{"authors":null,"categories":null,"content":" We organized a meeting between our lab and [Praxiling](http://www.praxiling.fr/?lang=en) a lab working in linguistic to explore potential collaborations between ecology and linguistic. Marie Chandelier and I gave a talk on our recent results on a content analysis we performed on wolf recovery in France and its coverage by newspapers. More soon on this front. ","date":1488196800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1488196800,"objectID":"9aa419d4c242b85e000d835ca0876fb3","permalink":"https://oliviergimenez.github.io/blog/praxiling/","publishdate":"2017-02-27T12:00:00Z","relpermalink":"/blog/praxiling/","section":"blog","summary":"We organized a meeting between our lab and [Praxiling](http://www.praxiling.fr/?lang=en) a lab working in linguistic to explore potential collaborations between ecology and linguistic. Marie Chandelier and I gave a talk on our recent results on a content analysis we performed on wolf recovery in France and its coverage by newspapers.","tags":["seminar","interdisciplinarity","linguistic"],"title":"Ecology meets linguistic","type":"blog"},{"authors":null,"categories":null,"content":"We had our first meeting for the project \u0026ldquo;Climate and management effects on COMmunity dynamics – Developing multi-species DEMOgraphy\u0026quot;funded by ANR. The website is operational and running, check it out here.\n","date":1487851200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1487851200,"objectID":"9f7bf3e86531d3fb613d80d82be2c36a","permalink":"https://oliviergimenez.github.io/blog/anr_democom_kickoff/","publishdate":"2017-02-23T12:00:00Z","relpermalink":"/blog/anr_democom_kickoff/","section":"blog","summary":"We had our first meeting for the project \u0026ldquo;Climate and management effects on COMmunity dynamics – Developing multi-species DEMOgraphy\u0026quot;funded by ANR. The website is operational and running, check it out here.","tags":["grant","project","multispecies","demography"],"title":"DEMOCOM kick-off meeting","type":"blog"},{"authors":null,"categories":null,"content":"We had our first meeting for the project \u0026ldquo;Scenarios for coexistence between men and dolphins on the Mediterranean French littoral\u0026rdquo; funded by Fondation de France. The website is operational and running, check it out here.\n","date":1485777600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1485777600,"objectID":"6b695cbd1c14e9f05b5227c22e9453ee","permalink":"https://oliviergimenez.github.io/blog/interact_kickoff/","publishdate":"2017-01-30T12:00:00Z","relpermalink":"/blog/interact_kickoff/","section":"blog","summary":"We had our first meeting for the project \u0026ldquo;Scenarios for coexistence between men and dolphins on the Mediterranean French littoral\u0026rdquo; funded by Fondation de France. The website is operational and running, check it out here.","tags":["grant","project","human-wildlife interactions","dolphins"],"title":"INTERACT kick-off meeting","type":"blog"},{"authors":null,"categories":null,"content":"We organised a 2-day workshop on the metrics of longevity used in human and animal demography. The idea is to explore potential bridges between the two fields. Next meeting in April. The group is led by S. Cubaynes, a former PhD student of mine, who now holds a lecturer position at Montpellier University.\n","date":1485432000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1485432000,"objectID":"c7ed4d141086ecde8239c7c06e49721f","permalink":"https://oliviergimenez.github.io/blog/working_group_demo/","publishdate":"2017-01-26T12:00:00Z","relpermalink":"/blog/working_group_demo/","section":"blog","summary":"We organised a 2-day workshop on the metrics of longevity used in human and animal demography. The idea is to explore potential bridges between the two fields. Next meeting in April.","tags":["working group","demography","R"],"title":"Working group on animal/human demography","type":"blog"},{"authors":null,"categories":null,"content":" Ian Renner from the University of Newcastle in Australia, who visited us back in 2015, is with us again for a short 3-week visit.\nWe resumed the work we started one year ago on developing a method to combine sources of information on species distribution. We also started a new project on model selection in occupancy and capture-recapture models using the LASSO.\n","date":1483272000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1483272000,"objectID":"6ab329f532b7cb9145eac4d27430b1fc","permalink":"https://oliviergimenez.github.io/blog/ian_visit2/","publishdate":"2017-01-01T12:00:00Z","relpermalink":"/blog/ian_visit2/","section":"blog","summary":" Ian Renner from the University of Newcastle in Australia, who visited us back in 2015, is with us again for a short 3-week visit.\n","tags":["visitor","species distribution models"],"title":"Another visit of Ian Renner","type":"blog"},{"authors":null,"categories":null,"content":"Blaise defended his PhD I co-supervised with Pierre-Yves Quenette from ONCFS (manuscript here). Blaise studied the attitudes of the public toward brown bear presence and provided sound estimates of abundance and distribution for the species in the Pyrenees.\nWe had stimulating discussions with the committee and F. Sarrazin, G. Chapron, C. Mounet, D. Grémillet and A. Molinari.\n","date":1481544000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1481544000,"objectID":"1a065145e2d1151b8ad7d03b7b0cc65c","permalink":"https://oliviergimenez.github.io/blog/blaise_defense/","publishdate":"2016-12-12T12:00:00Z","relpermalink":"/blog/blaise_defense/","section":"blog","summary":"Blaise defended his PhD I co-supervised with Pierre-Yves Quenette from ONCFS (manuscript here). Blaise studied the attitudes of the public toward brown bear presence and provided sound estimates of abundance and distribution for the species in the Pyrenees.\n","tags":["PhD","brown bear","social sciences"],"title":"Blaise defended his PhD on a socio-ecological approach to Brown bear management in the Pyrenees","type":"blog"},{"authors":null,"categories":null,"content":"Frederic Gosselin, Etienne Rivot and I organised a 2-day workshop on issues in fitting hierarchical models in ecology. I gave a talk on \u0026ldquo;Local minima and multistate capture-recapture models\u0026rdquo;, slides and R code available on GitHub.\nThe idea is to consider several problematic case studies and explore the issues using several computing platforms (R, Jags, Nimble, Stan, Admb). Next meeting in May.\n","date":1480334400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1480334400,"objectID":"73f675f4800f48b8e2e59c67f65baf19","permalink":"https://oliviergimenez.github.io/blog/working_group_hm/","publishdate":"2016-11-28T12:00:00Z","relpermalink":"/blog/working_group_hm/","section":"blog","summary":"Frederic Gosselin, Etienne Rivot and I organised a 2-day workshop on issues in fitting hierarchical models in ecology. I gave a talk on \u0026ldquo;Local minima and multistate capture-recapture models\u0026rdquo;, slides and R code available on GitHub.\n","tags":["working group","hierarchical models","R"],"title":"Workshop on issues in fitting hierarchical models in ecology","type":"blog"},{"authors":null,"categories":null,"content":"During the period 2013-2015, the NGO GIS3M led the GDEGeM project in which tons of data (on abundance and distribution among others) were collected on bottlenose dolphins.\nWe had a 1-day seminar gathering participants to the project, local authorities, lay persons who came to listen to the results of the projects. The talks can be seen here. I contributed by presenting our abundance estimates\nand our new project on the coexistence between dolphins and human activities ","date":1478779200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1478779200,"objectID":"c12dbd0aa2e6e11c1b858f924668bd91","permalink":"https://oliviergimenez.github.io/blog/gdegem_restitution/","publishdate":"2016-11-10T12:00:00Z","relpermalink":"/blog/gdegem_restitution/","section":"blog","summary":"During the period 2013-2015, the NGO GIS3M led the GDEGeM project in which tons of data (on abundance and distribution among others) were collected on bottlenose dolphins.\n","tags":["conference","dolphin","capture-recapture","social sciences"],"title":"GDEGeM workshop on bottlenose dolphin conservation in the North-Western Mediterranean Sea","type":"blog"},{"authors":null,"categories":null,"content":"We attended the conference organized by the French Ecological Society in Marseille. Our talk was \u0026lsquo;Conservation and management of large carnivores in France: A beneficial collaboration between ONCFS practitioners and CNRS researchers\u0026rsquo;.\nIt was a contribution to the Invited Symposium \u0026ldquo;Reciprocal benefits between practitioners, researchers and the biodiversity\u0026rdquo;. This was joint work with M. Guinot-Ghestem, E. Marboutin, Pierre-Yves Quenette, Laetitia Blanc, Sarah Cubaynes, Christophe Duchamp, Julie Louvrier, Lucile Marescot and Blaise Piédallu. The abstract is here.\n","date":1477396800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1477396800,"objectID":"fda79420301b8cb21db5ffbc6e9c8bae","permalink":"https://oliviergimenez.github.io/blog/sfe_2016/","publishdate":"2016-10-25T12:00:00Z","relpermalink":"/blog/sfe_2016/","section":"blog","summary":"We attended the conference organized by the French Ecological Society in Marseille. Our talk was \u0026lsquo;Conservation and management of large carnivores in France: A beneficial collaboration between ONCFS practitioners and CNRS researchers\u0026rsquo;.\n","tags":["conference","talk"],"title":"Talk on the science-management gap in Marseille","type":"blog"},{"authors":null,"categories":null,"content":" Raluca Bancila, who visited us back in 2015, is with us again for a short 2-week visit. We will resume a project we started one year ago on the evaluation of demographic parameters of two related species of spadefoot toads (Pelobates syriacus and P. fuscus) from the Danube Delta Biosphere Reserve, Romania, using capture-recapture models.\n","date":1475323200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1475323200,"objectID":"bb8406dcfe9c4eb8089202e631817aeb","permalink":"https://oliviergimenez.github.io/blog/raluca_visit2/","publishdate":"2016-10-01T12:00:00Z","relpermalink":"/blog/raluca_visit2/","section":"blog","summary":"Raluca Bancila, who visited us back in 2015, is with us again for a short 2-week visit. We will resume a project we started one year ago on the evaluation of demographic parameters of two related species of spadefoot toads (Pelobates syriacus and P.","tags":["visitor","capture-recapture"],"title":"Another visit of Raluca Bancila","type":"blog"},{"authors":null,"categories":null,"content":" Laura Cowen from the Department of Math and Stat at University of Victoria (Canada) will be on a sabbatical with us for 3 months. We will be working on the applications of hidden Markov models to capture-recapture data.\n","date":1472731200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1472731200,"objectID":"553ac55f0a6dde441c7017b62348caef","permalink":"https://oliviergimenez.github.io/blog/laura_cowen/","publishdate":"2016-09-01T12:00:00Z","relpermalink":"/blog/laura_cowen/","section":"blog","summary":"Laura Cowen from the Department of Math and Stat at University of Victoria (Canada) will be on a sabbatical with us for 3 months. We will be working on the applications of hidden Markov models to capture-recapture data.","tags":["visitor","hmm"],"title":"Laura Cowen's sabbatical","type":"blog"},{"authors":null,"categories":null,"content":"This is a free adaptation of two (very) clever analyses made by others:\nThe Star Wars Social Network by Evelina Gabasov in which program F# was mostly used to analyse the Star wars social networks\nAnalyzing networks of characters in \u0026lsquo;Love Actually\u0026rsquo; by David Robinson in which R was used to analyse the links between the characters of the movie Love Actually.\nThe aim here is to try and reproduce Evelina\u0026rsquo;s analysis using R only, using David\u0026rsquo;s contribution plus several tweaks I found here and there on the internet. The R code and data are available on my GitHub page.\nDisclaimer: The original blog posts are awesome and full of relevant details, check them out! My objective here was to teach myself how to manipulate data using trendy R packages and do some network analyses. Some comments below have been copied and pasted from these blogs, the credits entirely go to the authors Evelina and David. Last but not least, my code comes with mistakes probably.\nRead and format data First, read in data. I found the movie script in doc format here, which I converted in txt format for convenience. Then, apply various treatments to have the data ready for analysis. I use the old school way for modifying the original dataframe. Piping would have made the code more readable, but I do not feel confident with this approach yet.\n# load convenient packages library(dplyr) library(stringr) library(tidyr) # read file line by line raw \u0026lt;- readLines(\u0026quot;attack-of-the-clones.txt\u0026quot;) # create data frame lines \u0026lt;- data_frame(raw = raw) # get rid of leading and trailing white spaces # http://stackoverflow.com/questions/2261079/how-to-trim-leading-and-trailing-whitespace-in-r trim \u0026lt;- function (x) gsub(\u0026quot;^\\\\s+|\\\\s+$\u0026quot;, \u0026quot;\u0026quot;, x) lines \u0026lt;- mutate(lines,raw=trim(raw)) # get rid of the empty lines lines2 \u0026lt;- filter(lines, raw != \u0026quot;\u0026quot;) # detect scenes: begin by EXT. or INT. lines3 \u0026lt;- mutate(lines2, is_scene = str_detect(raw, \u0026quot;T.\u0026quot;),scene = cumsum(is_scene)) # drop lines that start with EXT. or INT. lines4 \u0026lt;- filter(lines3,!is_scene) # distinguish characters from what they say lines5 \u0026lt;- separate(lines4, raw, c(\u0026quot;speaker\u0026quot;, \u0026quot;dialogue\u0026quot;), sep = \u0026quot;:\u0026quot;, fill = \u0026quot;left\u0026quot;,extra='drop') # read in aliases (from Evelina's post) aliases \u0026lt;- read.table('aliases.csv',sep=',',header=T,colClasses = \u0026quot;character\u0026quot;) aliases$Alias ## [1] \u0026quot;BEN\u0026quot; \u0026quot;SEE-THREEPIO\u0026quot; \u0026quot;THREEPIO\u0026quot; \u0026quot;ARTOO-DETOO\u0026quot; ## [5] \u0026quot;ARTOO\u0026quot; \u0026quot;PALPATINE\u0026quot; \u0026quot;DARTH SIDIOUS\u0026quot; \u0026quot;BAIL\u0026quot; ## [9] \u0026quot;MACE\u0026quot; \u0026quot;WINDU\u0026quot; \u0026quot;MACE-WINDU\u0026quot; \u0026quot;NUTE\u0026quot; ## [13] \u0026quot;AUNT BERU\u0026quot; \u0026quot;DOOKU\u0026quot; \u0026quot;BOBA\u0026quot; \u0026quot;JANGO\u0026quot; ## [17] \u0026quot;PANAKA\u0026quot; \u0026quot;NUTE\u0026quot; \u0026quot;KI-ADI\u0026quot; \u0026quot;BIBBLE\u0026quot; ## [21] \u0026quot;BIB\u0026quot; \u0026quot;CHEWIE\u0026quot; \u0026quot;VADER\u0026quot; aliases$Name ## [1] \u0026quot;OBI-WAN\u0026quot; \u0026quot;C-3PO\u0026quot; \u0026quot;C-3PO\u0026quot; \u0026quot;R2-D2\u0026quot; ## [5] \u0026quot;R2-D2\u0026quot; \u0026quot;EMPEROR\u0026quot; \u0026quot;EMPEROR\u0026quot; \u0026quot;BAIL ORGANA\u0026quot; ## [9] \u0026quot;MACE WINDU\u0026quot; \u0026quot;MACE WINDU\u0026quot; \u0026quot;MACE WINDU\u0026quot; \u0026quot;NUTE GUNRAY\u0026quot; ## [13] \u0026quot;BERU\u0026quot; \u0026quot;COUNT DOOKU\u0026quot; \u0026quot;BOBA FETT\u0026quot; \u0026quot;JANGO FETT\u0026quot; ## [17] \u0026quot;CAPTAIN PANAKA\u0026quot; \u0026quot;NUTE GUNRAY\u0026quot; \u0026quot;KI-ADI-MUNDI\u0026quot; \u0026quot;SIO BIBBLE\u0026quot; ## [21] \u0026quot;BIB FORTUNA\u0026quot; \u0026quot;CHEWBACCA\u0026quot; \u0026quot;DARTH VADER\u0026quot; # assign unique name to characters # http://stackoverflow.com/questions/28593265/is-there-a-function-like-switch-which-works-inside-of-dplyrmutate multipleReplace \u0026lt;- function(x, what, by) { stopifnot(length(what)==length(by)) ind \u0026lt;- match(x, what) ifelse(is.na(ind),x,by[ind]) } lines6 \u0026lt;- mutate(lines5,speaker=multipleReplace(speaker,what=aliases$Alias,by=aliases$Name)) # read in actual names (from Evelina's post) actual.names \u0026lt;- read.csv('characters.csv',header=F,colClasses = \u0026quot;character\u0026quot;) actual.names \u0026lt;- c(as.matrix(actual.names)) # filter out non-characters lines7 \u0026lt;- filter(lines6,speaker %in% actual.names) # group by scene lines8 \u0026lt;- group_by(lines7, scene, line = cumsum(!is.na(speaker))) lines9 \u0026lt;- summarize(lines8, speaker = speaker[1], dialogue = str_c(dialogue, collapse = \u0026quot; \u0026quot;)) # Count the lines-per-scene-per-character # Turn the result into a binary speaker-by-scene matrix by_speaker_scene \u0026lt;- count(lines9, scene, speaker) by_speaker_scene ## # A tibble: 447 x 3 ## # Groups: scene [321] ## scene speaker n ## \u0026lt;int\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;int\u0026gt; ## 1 26 PADME 1 ## 2 27 PADME 1 ## 3 29 PADME 1 ## 4 48 PADME 1 ## 5 50 PADME 2 ## 6 66 MACE WINDU 1 ## 7 67 MACE WINDU 1 ## 8 69 YODA 1 ## 9 70 MACE WINDU 1 ## 10 74 YODA 1 ## # ... with 437 more rows library(reshape2) speaker_scene_matrix \u0026lt;-acast(by_speaker_scene , speaker ~ scene, fun.aggregate = length) dim(speaker_scene_matrix) ## [1] 19 321 Analyses Hierarchical clustering norm \u0026lt;- speaker_scene_matrix / rowSums(speaker_scene_matrix) h \u0026lt;- hclust(dist(norm, method = \u0026quot;manhattan\u0026quot;)) plot(h) Timeline Use tree to give an ordering that puts similar characters close together\nordering \u0026lt;- h$labels[h$order] ordering ## [1] \u0026quot;MACE WINDU\u0026quot; \u0026quot;YODA\u0026quot; \u0026quot;SHMI\u0026quot; \u0026quot;QUI-GON\u0026quot; \u0026quot;PLO KOON\u0026quot; ## [6] \u0026quot;LAMA SU\u0026quot; \u0026quot;OBI-WAN\u0026quot; \u0026quot;BAIL ORGANA\u0026quot; \u0026quot;JAR JAR\u0026quot; \u0026quot;POGGLE\u0026quot; ## [11] \u0026quot;ANAKIN\u0026quot; \u0026quot;PADME\u0026quot; \u0026quot;CLIEGG\u0026quot; \u0026quot;BERU\u0026quot; \u0026quot;OWEN\u0026quot; ## [16] \u0026quot;SIO BIBBLE\u0026quot; \u0026quot;RUWEE\u0026quot; \u0026quot;JOBAL\u0026quot; \u0026quot;SOLA\u0026quot; This ordering can be used to make other graphs more informative. For instance, we can visualize a timeline of all scenes:\nscenes \u0026lt;- filter(by_speaker_scene, n() \u0026gt; 1) # scenes with \u0026gt; 1 character scenes2 \u0026lt;- ungroup(scenes) scenes3 \u0026lt;- mutate(scenes2, scene = as.numeric(factor(scene)), character = factor(speaker, levels = ordering)) library(ggplot2) ggplot(scenes3, aes(scene, character)) + geom_point() + geom_path(aes(group = scene)) Create a cooccurence matrix (see here) containing how many times two characters share scenes\ncooccur \u0026lt;- speaker_scene_matrix %*% t(speaker_scene_matrix) heatmap(cooccur) Social network analyses Graphical representation of the network Here the nodes represent characters in the movies. The characters are connected by a link if they both speak in the same scene. And the more the characters speak together, the thicker the link between them.\nlibrary(igraph) g \u0026lt;- graph.adjacency(cooccur, weighted = TRUE, mode = \u0026quot;undirected\u0026quot;, diag = FALSE) plot(g, edge.width = E(g)$weight) Compute standard network features, degree and betweeness.\ndegree(g) ## ANAKIN BAIL ORGANA BERU CLIEGG JAR JAR JOBAL ## 12 1 4 4 4 4 ## LAMA SU MACE WINDU OBI-WAN OWEN PADME PLO KOON ## 1 5 6 4 12 0 ## POGGLE QUI-GON RUWEE SHMI SIO BIBBLE SOLA ## 1 1 4 1 0 4 ## YODA ## 4 betweenness(g) ## ANAKIN BAIL ORGANA BERU CLIEGG JAR JAR JOBAL ## 42.600000 0.000000 1.750000 0.500000 22.000000 0.000000 ## LAMA SU MACE WINDU OBI-WAN OWEN PADME PLO KOON ## 0.000000 18.366667 15.000000 5.250000 55.133333 0.000000 ## POGGLE QUI-GON RUWEE SHMI SIO BIBBLE SOLA ## 0.000000 0.000000 0.700000 0.000000 0.000000 5.000000 ## YODA ## 3.366667 To get a nicer representation of the network, see here and the formating from igraph to d3Network. Below is the code you’d need:\nlibrary(d3Network) library(networkD3) sg \u0026lt;- simplify(g) df \u0026lt;- get.edgelist(g, names=TRUE) df \u0026lt;- as.data.frame(df) colnames(df) \u0026lt;- c('source', 'target') df$value \u0026lt;- rep(1, nrow(df)) # get communities fc \u0026lt;- fastgreedy.community(g) com \u0026lt;- membership(fc) node.info \u0026lt;- data.frame(name=names(com), group=as.vector(com)) links \u0026lt;- data.frame(source=match(df$source, node.info$name)-1,target=match(df$target, node.info$name)-1,value=df$value) forceNetwork(Links = links, Nodes = node.info,Source = \u0026quot;source\u0026quot;, Target = \u0026quot;target\u0026quot;,Value = \u0026quot;value\u0026quot;, NodeID = \u0026quot;name\u0026quot;,Group = \u0026quot;group\u0026quot;, opacity = 1, opacityNoHover=1) The nodes represent characters in the movies. The characters are connected by a link if they both speak in the same scene. The colors are for groups obtained by some algorithms.\n","date":1470571200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1470571200,"objectID":"1d35aafc874b8412f4d1332334926076","permalink":"https://oliviergimenez.github.io/blog/starwars_network/","publishdate":"2016-08-07T12:00:00Z","relpermalink":"/blog/starwars_network/","section":"blog","summary":"This is a free adaptation of two (very) clever analyses made by others:\nThe Star Wars Social Network by Evelina Gabasov in which program F# was mostly used to analyse the Star wars social networks","tags":["star wars","social networks","R","rstats"],"title":"Analysing the social Star Wars network in The Attack of the Clones with R","type":"blog"},{"authors":null,"categories":null,"content":"Wow, another good news, our project \u0026ldquo;Climate and management effects on COMmunity dynamics – Developing multi-species DEMOgraphy\u0026rdquo; will be funded by ANR!\nI submitted it in April for the third time. I originally submitted it (just once) to the ERC back in 2011, it was rejected. Over the years, we received constructive comments from referees that helped improving the project, and made it more solid.\nIn DEMOCOM, we will develop an integrated statistical framework for estimation and inference about community dynamics via the development of a multispecies demography, bringing together concepts, predictions and tools from both fields of demography and community ecology. We will apply this new framework to case studies on fishes, mammals and birds. The methods will be made available through R packages.\nThis is a collaborative project with many partners involved, and people to be hired (stay tuned). The project will start early 2017, more soon.\n","date":1469102400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1469102400,"objectID":"dd87e3141b90b50c723014e96cde02cf","permalink":"https://oliviergimenez.github.io/blog/anr_democom/","publishdate":"2016-07-21T12:00:00Z","relpermalink":"/blog/anr_democom/","section":"blog","summary":"Wow, another good news, our project \u0026ldquo;Climate and management effects on COMmunity dynamics – Developing multi-species DEMOgraphy\u0026rdquo; will be funded by ANR!\n","tags":["grant","project","multispecies","demography"],"title":"Research grant on multispecies demography","type":"blog"},{"authors":null,"categories":null,"content":"I decided to teach myself how to do proper reproducible research. Many reasons to that: save time on the mid/long term, make my analyses open and criticizable, share with others, \u0026hellip;\nThere are tons of resources on the web. I use the tutorials cooked by guru Karl Broman. For people like me who learned S-plus, he has some advice to switch to modern tools. Among others, I try and use more and more RStudio in connection with KnitR and R Markdown and share my codes on GitHub. I am more a Fortran guy, but Rcpp makes me wonder about C++. Version control, piping and the tidy universe are not yet entirely familiar to me, I’m making baby steps (or grandpa steps I should say). Let’s see how it goes.\n","date":1468152000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1468152000,"objectID":"f18772904a9680d4411f3d6c63270c3a","permalink":"https://oliviergimenez.github.io/blog/reproducible_research/","publishdate":"2016-07-10T12:00:00Z","relpermalink":"/blog/reproducible_research/","section":"blog","summary":"I decided to teach myself how to do proper reproducible research. Many reasons to that: save time on the mid/long term, make my analyses open and criticizable, share with others, \u0026hellip;\n","tags":["programming","reproducible research","R"],"title":"Self teaching reproducible research","type":"blog"},{"authors":null,"categories":null,"content":"I attended the International Statistical Ecology Conference in Seattle. I don’t go often to conferences, but I do my best not to miss that one.\nThis year, clearly, hidden Markov models (HMM) were on the lips of everyone. Well, it was cool to realize that I wasn\u0026rsquo;t too \u0026ldquo;has been\u0026rdquo; by giving a 1-day workshop on fitting occupancy models with HMM and program E-SURGE developed in our team.\nI also gave a talk on the use of PCA to deal with many correlated covariated in capture-recapture models.\nI also co-authored a talk that Ian Renner (see below) gave on combining sources of information on species distribution.\n","date":1467374400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1467374400,"objectID":"e78d51d764ab0122d73733e22ea564fd","permalink":"https://oliviergimenez.github.io/blog/isec_2016/","publishdate":"2016-07-01T12:00:00Z","relpermalink":"/blog/isec_2016/","section":"blog","summary":"I attended the International Statistical Ecology Conference in Seattle. I don’t go often to conferences, but I do my best not to miss that one.\n","tags":["conference","isec"],"title":"International Statistical Ecology Conference in Seattle","type":"blog"},{"authors":null,"categories":null,"content":"Great news! Our project on \u0026ldquo;Scenarios for coexistence between men and dolphins on the Mediterranean French littoral\u0026rdquo; will be funded by Fondation de France.\nIn this project, we will propose management strategies for human activities and populations of bottlenose dolphins to coexist. Building on the GDEGeM project, we will adopt an ecological approach to infer the spatio-temporal distribution of dolphins and their interactions with human activities, and a sociological approach to explore the attitudes and perceptions of local actors towards dolphins. We will resort to participatory modeling to build, test and use a model to assess management strategies collectively.\nIn passing, I kind of liked the selection procedure: we first submitted a written application, then were asked to come for an interview. I could answer the questions of the committee and have a direct feeling of what they were thinking of our project.\nWe will start this project officially by the end of the year, more soon.\n","date":1464782400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1464782400,"objectID":"dc566a6c89404b085b6b548390853b49","permalink":"https://oliviergimenez.github.io/blog/interact/","publishdate":"2016-06-01T12:00:00Z","relpermalink":"/blog/interact/","section":"blog","summary":"Great news! Our project on \u0026ldquo;Scenarios for coexistence between men and dolphins on the Mediterranean French littoral\u0026rdquo; will be funded by Fondation de France.\n","tags":["grant","human-wildlife interactions","dolphins"],"title":"Research grant on human-dolphin interactions","type":"blog"},{"authors":null,"categories":null,"content":"My post-doc Sarah Cubaynes and I went to visit our colleague Jon Aars in Tromso, Norway. We are working together on the demography of polar bears using an amazing dataset the Norwegian Polar Institute has been collecting for decades. It was also the opportunity to catch up with Nigel (Gilles) Yoccoz who\u0026rsquo;s being very helpful with this project. Jon took us fishing; I was luckier than with fly-fishing ;-)\n","date":1464609600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1464609600,"objectID":"0750e7a68d62c6e8acc925ff36559a0b","permalink":"https://oliviergimenez.github.io/blog/polar_bear/","publishdate":"2016-05-30T12:00:00Z","relpermalink":"/blog/polar_bear/","section":"blog","summary":"My post-doc Sarah Cubaynes and I went to visit our colleague Jon Aars in Tromso, Norway. We are working together on the demography of polar bears using an amazing dataset the Norwegian Polar Institute has been collecting for decades.","tags":["trip","polar bear","Norway","demography"],"title":"Visiting Tromso, polar bear demography","type":"blog"},{"authors":null,"categories":null,"content":"We organised the second conference of the GdR EcoStat. We spent two lovely days in Montpellier talking about ecological statistics with the hundred people who attended. The talks are available here. If you\u0026rsquo;d like to know more about our acitivities in the GdR, check out our website.\n","date":1464264000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1464264000,"objectID":"7aa7d56683dd1e3f5fa9c96a70e8f8fb","permalink":"https://oliviergimenez.github.io/blog/gdr_ecostat_journee2/","publishdate":"2016-05-26T12:00:00Z","relpermalink":"/blog/gdr_ecostat_journee2/","section":"blog","summary":"We organised the second conference of the GdR EcoStat. We spent two lovely days in Montpellier talking about ecological statistics with the hundred people who attended. The talks are available here.","tags":["conference","statistical ecology"],"title":"GDR EcoStat meeting","type":"blog"},{"authors":null,"categories":null,"content":"Among the papers we published lately, I’d like to emphasize two contributions on the conservation of brown bears with some eminent colleagues:\nKaramanlidis, A.A., M. de Gabriel Hernando, L. Krambokoukis, O. Gimenez (2015). Evidence of a large carnivore population recovery: counting bears in Greece. Journal for Nature Conservation. This is a joint venture with Alex and the Greek Arcturos NGO in which we provide the first reliable abundance estimates for brown bears in Greece.\nBischof, R., H. Broseth, O. Gimenez (2015). Wildlife in a politically divided world: insularism inflates estimates of brown bear abundance. Conservation Letters. Richard had this very clever idea to use spatially-explicit capture-recapture models to assess to what extent abundance estimates are affected by double-counting animals that move across borders - see here for a nice summary of our work.\n","date":1464177600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1464177600,"objectID":"771a611ca4cfc9f8432248d57cd4c544","permalink":"https://oliviergimenez.github.io/blog/papers_bears/","publishdate":"2016-05-25T12:00:00Z","relpermalink":"/blog/papers_bears/","section":"blog","summary":"Among the papers we published lately, I’d like to emphasize two contributions on the conservation of brown bears with some eminent colleagues:\nKaramanlidis, A.A., M. de Gabriel Hernando, L. Krambokoukis, O.","tags":["paper","brown bear"],"title":"Focus on two Brown bear papers","type":"blog"},{"authors":null,"categories":null,"content":"","date":1457784000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1457784000,"objectID":"d51521748827ddfe814f7171510ff75f","permalink":"https://oliviergimenez.github.io/blog/fishing2016/","publishdate":"2016-03-12T12:00:00Z","relpermalink":"/blog/fishing2016/","section":"blog","summary":"","tags":["fishing"],"title":"Fishing season is open ;-)","type":"blog"},{"authors":null,"categories":null,"content":"The GIS3M held an international workshop in Marseille on the conservation of Bottlenose dolphin conservation and monitoring in the North-Western Mediterranean Sea. This was the opportunity to discuss the results of the GDEGeM project on the same topic. I mostly contributed by analysing the tons of photo-IDs that were collected during the program to estimate abundance and analyze the social structure. The slides of my talk:\n","date":1448971200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1448971200,"objectID":"fba7bca8edf18b8cd94ef1e95d57fc36","permalink":"https://oliviergimenez.github.io/blog/gis3m/","publishdate":"2015-12-01T12:00:00Z","relpermalink":"/blog/gis3m/","section":"blog","summary":"The GIS3M held an international workshop in Marseille on the conservation of Bottlenose dolphin conservation and monitoring in the North-Western Mediterranean Sea. This was the opportunity to discuss the results of the GDEGeM project on the same topic.","tags":["conference","dolphins","capture-recapture"],"title":"International workshop on Bottlenose dolphin conservation and monitoring","type":"blog"},{"authors":null,"categories":null,"content":"Jean-Dominique Lebreton went on retirement, and we could not let him go without organizing something in his honor. Jean-Do had a tremendous influence in the field of statistical ecology with important contributions in (among others) multivariate analyses, software developments, population dynamics, and capture-recapture methods. On a personal side, I would simply not be here in science without him. Many friends and colleagues came or sent a message for the occasion. These interventions were filmed, see here and enjoy.\n","date":1447416000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1447416000,"objectID":"2d8ee50aa32bdbc468e8cc8cca149c31","permalink":"https://oliviergimenez.github.io/blog/jdl/","publishdate":"2015-11-13T12:00:00Z","relpermalink":"/blog/jdl/","section":"blog","summary":"Jean-Dominique Lebreton went on retirement, and we could not let him go without organizing something in his honor. Jean-Do had a tremendous influence in the field of statistical ecology with important contributions in (among others) multivariate analyses, software developments, population dynamics, and capture-recapture methods.","tags":["conference","statistical ecology"],"title":"Tribute to Jean-Dominique Lebreton","type":"blog"},{"authors":null,"categories":null,"content":"Taking the opportunity of Ian Renner’s visit, we organized thanks to the GdR EcoStat a conference on the recent advances in the modeling of species distribution. We were lucky to have the cream of researchers in the domain with Julien Papaïx, Joseph Chipperfield, Ian Renner, Joern Pagel, Bob O’Hara, Laura Pollock and Damaris Zurell. The talks are available here.\n","date":1446552000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1446552000,"objectID":"f56c5cadeaa24f1f8ddc4607b1c435db","permalink":"https://oliviergimenez.github.io/blog/sdm_day/","publishdate":"2015-11-03T12:00:00Z","relpermalink":"/blog/sdm_day/","section":"blog","summary":"Taking the opportunity of Ian Renner’s visit, we organized thanks to the GdR EcoStat a conference on the recent advances in the modeling of species distribution. We were lucky to have the cream of researchers in the domain with Julien Papaïx, Joseph Chipperfield, Ian Renner, Joern Pagel, Bob O’Hara, Laura Pollock and Damaris Zurell.","tags":["conference","talk"],"title":"Species distribution model day","type":"blog"},{"authors":null,"categories":null,"content":"Ian Renner from the University of Newcastle in Australia spent some time with us while on a sabbatical. We worked together on developing a method to combine sources of information on species distribution. Ian brought his sound expertise on the Lasso for covariate selection and point processes that provide a nice integrated framework. It was fun to program in R in parallel. I wish I could do that more often. We even managed (credit goes to Ian) to present something at the International Statistical Ecology Conference.\nOn a personal side, Ian has become a friend, his kindness and interest in other people are amazing.\n","date":1445860800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1445860800,"objectID":"2748bb81619a0a5be9cc06b685f03f1e","permalink":"https://oliviergimenez.github.io/blog/ian_first_visit/","publishdate":"2015-10-26T12:00:00Z","relpermalink":"/blog/ian_first_visit/","section":"blog","summary":"Ian Renner from the University of Newcastle in Australia spent some time with us while on a sabbatical. We worked together on developing a method to combine sources of information on species distribution.","tags":["visitor"],"title":"Ian Renner's visit","type":"blog"},{"authors":null,"categories":null,"content":"I had the visit of Raluca Bancila a research scientist from Romania. We investigated spatio-temporal environmental variation in and individual heterogeneity on emigration/immigration rates of the troglophile harvestmen species Paranemastoma silli with capture-recapture models. We used two P. silli populations inhabiting two caves, Closani and Lazului Caves, located in the Mehedinti Mountains in southwestern Romania. Yes, we do not only work on big emblematic mammal species ;-) On a personal side, Raluca is always smiling and laughing – ‘la banane’ like we say in French, it was so good to have her in the team. I’m looking forward to hosting her again!\n","date":1443700800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1443700800,"objectID":"061671e16d815f974e6aa2b17ee6d6e4","permalink":"https://oliviergimenez.github.io/blog/raluca_visit1/","publishdate":"2015-10-01T12:00:00Z","relpermalink":"/blog/raluca_visit1/","section":"blog","summary":"I had the visit of Raluca Bancila a research scientist from Romania. We investigated spatio-temporal environmental variation in and individual heterogeneity on emigration/immigration rates of the troglophile harvestmen species Paranemastoma silli with capture-recapture models.","tags":["visitor","individual heterogeneity","capture-recapture"],"title":"Raluca Bancila's visit","type":"blog"},{"authors":null,"categories":null,"content":"We visited our colleagues from Dolphin Biology and Conservation (Giovanni Bearzi, Silvia Bonizzoni, Nina Santostasi ‎and‎‎ Eva Greene). Lots of discussions on future projects about the estimation of striped, bottlenose and common dolphins abundance, their distribution in the gulf of Corinth (Greece) and their interactions with fishermen. Nina will be visiting us in October-March to work on the abundance project. We also went out for a survey and spotted several groups of striped dolphins, and two common dolphins. Thanks to Giovanni and Silvia for their hospitality and kindness. The beginning of a new adventure in Greece!\n","date":1439380800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1439380800,"objectID":"84c4367d1877f224b9c890e02ddc0a7d","permalink":"https://oliviergimenez.github.io/blog/galaxidi/","publishdate":"2015-08-12T12:00:00Z","relpermalink":"/blog/galaxidi/","section":"blog","summary":"We visited our colleagues from Dolphin Biology and Conservation (Giovanni Bearzi, Silvia Bonizzoni, Nina Santostasi ‎and‎‎ Eva Greene). Lots of discussions on future projects about the estimation of striped, bottlenose and common dolphins abundance, their distribution in the gulf of Corinth (Greece) and their interactions with fishermen.","tags":["trip"],"title":"Dolphins, gulf of Corinth, Greece","type":"blog"},{"authors":null,"categories":null,"content":"I attended the International Congress on Biological Conservation that was held in my home town Montpellier. The students of our team were on fire and talked about their exciting work. Gilles presented a bio-economic model to study the population dynamics of Asian captive elephants, a nice example of how to explicitly incorporate interactions between captive and wild populations in species management.\nJulie talked about her use of opportunistic data to map the colonisation of the wolf in France; this is a neat example of how to generate non-detections by carefully assessing the observation efforts to use dynamic occupancy models.\nAnne-Sophie presented an application of Poisson point process modeling to map the presence of brown bear presence in Greece using citizen science data with possible observer bias.\nBlaise presented his results on a sociological survey to assess the attitudes of local residents towards the presence of Brown bears in the Pyrenees.\nLast, Laetitia had a poster on the use of a spatially explicit individual-based model to explore the best strategies (reintroductions and / or corridors) to restore the population of lynx in the Vosges mountains.\nI gave a talk on a review of the methods to analyse citizen science data. This was part of a symposium (nicely summarised by Muki Haklay here) organised by Karine Princé and Wes Hochachka. Stay tuned, a paper is on its way!\nOverall, the program of this conference was dense, the talks and discussions between sessions were stimulating, I’m looking forward to attending the next one!\n","date":1438516800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1438516800,"objectID":"bce30cc635c46d3c8094f60207d331a6","permalink":"https://oliviergimenez.github.io/blog/iccb2014/","publishdate":"2015-08-02T12:00:00Z","relpermalink":"/blog/iccb2014/","section":"blog","summary":"I attended the International Congress on Biological Conservation that was held in my home town Montpellier. The students of our team were on fire and talked about their exciting work. Gilles presented a bio-economic model to study the population dynamics of Asian captive elephants, a nice example of how to explicitly incorporate interactions between captive and wild populations in species management.","tags":["conference","citizen science"],"title":"International Congress on Biological Conservation","type":"blog"},{"authors":null,"categories":null,"content":"Just submitted a big proposal to the Biodiversa call. It was great to write this project with Hervé Fritz who, without a doubt, knows how to write grant applications! If successful, we\u0026rsquo;ll be working on the social and ecological dimensions of connectivity in lynx habitats. This is joint work with French, Swiss, German, Norwegian and Swedish colleagues. More soon hopefully.\nUpdate: On May 2016, I received the news that our project was rejected. Big disappointment. But hey, that\u0026rsquo;s life, and this project is awesome, we\u0026rsquo;ll find other ways to implement it!\n","date":1437393600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1437393600,"objectID":"f05ec51112367326998e453bb2956fd3","permalink":"https://oliviergimenez.github.io/blog/biodiversa/","publishdate":"2015-07-20T12:00:00Z","relpermalink":"/blog/biodiversa/","section":"blog","summary":"Just submitted a big proposal to the Biodiversa call. It was great to write this project with Hervé Fritz who, without a doubt, knows how to write grant applications! If successful, we\u0026rsquo;ll be working on the social and ecological dimensions of connectivity in lynx habitats.","tags":["grant","lynx","socio-ecological approach"],"title":"Biodiversa grant submission","type":"blog"},{"authors":null,"categories":null,"content":"I had the visit of Sarah Benhaim and Lucile Marescot (a former PhD student of mine), both post-doc at the Leibniz Institute in Berlin supervised by Stephanie Kramer-Schadt. It was an exciting week that we spent discussing and analyzing data on the spotted hyenas from the Serengeti National Park with questions at the interface of demography, sociality and epidemiology. This is joint work with Heribert Hofer and Marion East.\n","date":1435752000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1435752000,"objectID":"cc949ea6a0eb71acf6c052dd3a34ae49","permalink":"https://oliviergimenez.github.io/blog/visit_from_izw/","publishdate":"2015-07-01T12:00:00Z","relpermalink":"/blog/visit_from_izw/","section":"blog","summary":"I had the visit of Sarah Benhaim and Lucile Marescot (a former PhD student of mine), both post-doc at the Leibniz Institute in Berlin supervised by Stephanie Kramer-Schadt. It was an exciting week that we spent discussing and analyzing data on the spotted hyenas from the Serengeti National Park with questions at the interface of demography, sociality and epidemiology.","tags":["visitor"],"title":"Visitors from Berlin","type":"blog"},{"authors":null,"categories":null,"content":"I\u0026rsquo;m so delighted to announce that Julie will start a PhD in October with me, jointly working with Eric Marboutin and Christophe Duchamp from ONCFS. There will be also collaborations with John Linnell from NINA. She\u0026rsquo;ll be studying the various ways to assess the distribution of large carnivores and methods to optimise the associated monitoring protocols.\n","date":1435233600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1435233600,"objectID":"fb37f572f0b4a98b8aa2ad6a3ab38784","permalink":"https://oliviergimenez.github.io/blog/grant_julie/","publishdate":"2015-06-25T12:00:00Z","relpermalink":"/blog/grant_julie/","section":"blog","summary":"I\u0026rsquo;m so delighted to announce that Julie will start a PhD in October with me, jointly working with Eric Marboutin and Christophe Duchamp from ONCFS. There will be also collaborations with John Linnell from NINA.","tags":["student","large carnivores"],"title":"New PhD student on large carnivore monitoring","type":"blog"},{"authors":null,"categories":null,"content":"My 2 master students finished their internships. I was so lucky to have them and work with them. Iago worked on the estimation of abundance using bilateral photo-identification capture-recapture data, with an application to bottlenose dolphins. He made a very neat contribution to the field by greatly improving existing algorithms (here and here) to fit capture-recapture models to these non-standard data. Julie worked on inferring the distribution of wolves in France from the beginning of its recolonization in the early 90’s. Her main contribution was to show that by carefully describing observers’ characteristics and properly quantify the prospection effort, citizen science data can be used to map the range dynamics of species with large dispersal distances and home ranges. Iago will spend next year improving his (already largely above the average) skills in informatics, AI, programming, agent-based modeling\u0026hellip; Hopefully, we’ll work again together. Julie is applying for PhD grants, more soon.\n","date":1433160000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1433160000,"objectID":"573248aab9a36227564ce868eb047d2b","permalink":"https://oliviergimenez.github.io/blog/master_2015/","publishdate":"2015-06-01T12:00:00Z","relpermalink":"/blog/master_2015/","section":"blog","summary":"My 2 master students finished their internships. I was so lucky to have them and work with them. Iago worked on the estimation of abundance using bilateral photo-identification capture-recapture data, with an application to bottlenose dolphins.","tags":["student"],"title":"Awesome master students","type":"blog"},{"authors":null,"categories":null,"content":"We had our annual department day; for the occasion, we invited Anne-Caroline Prévot. Anne-Caro is having a fascinating scientific pathway, as she started with a PhD in bird population dynamics and now she’s working on topics requiring skills in environmental psychology. Google her, you’ll see for yourself, her work is awesome. For this special day, we opted for a particular format, with less science than usual, the idea being to show what’s going on behind the scenes. We had a roundtable on the link between research, managers and citizens. We also had several short unorthodox talks where we tried to analyse why some of our papers/projects were successful, but also why we failed at being read or funded. Interestingly, Rob Salguero-Gomez, a colleague of ours, wrote a nice piece on the topic here. Last, we gathered several persons with various professional profiles so that our students could ask them all the questions they wanted. Overall, I think it went amazingly well, and all the attendees found something to take home with them.\n","date":1432209600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1432209600,"objectID":"e799c8f7876ce05f0bef914af399b852","permalink":"https://oliviergimenez.github.io/blog/department_meeting_2015/","publishdate":"2015-05-21T12:00:00Z","relpermalink":"/blog/department_meeting_2015/","section":"blog","summary":"We had our annual department day; for the occasion, we invited Anne-Caroline Prévot. Anne-Caro is having a fascinating scientific pathway, as she started with a PhD in bird population dynamics and now she’s working on topics requiring skills in environmental psychology.","tags":["CEFE","meeting"],"title":"Biodiversity and conservation department meeting","type":"blog"},{"authors":null,"categories":null,"content":"As one of our always excellent department seminar, we had the chance to welcome Heath Smith from University of Washington, Center for Conservation Biology. Heath, together with his dog Chester with whom I became friend during the talk, is working with detection dogs on all sorts of important questions and organisms in conservation biology. Heath then headed to the Pyrénées with Nathalie Espuno to meet with our colleagues from the ONCFS bear (dream)team with whom we\u0026rsquo;re working on Blaise Piédallu\u0026rsquo;s PhD. They\u0026rsquo;re gonna try this technique there, no doubt the results will be of interest for our work on large carnivores!\n","date":1431518400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1431518400,"objectID":"cb9da998721be2e3cf8a1293a9d5f294","permalink":"https://oliviergimenez.github.io/blog/seminar_detection_dogs/","publishdate":"2015-05-13T12:00:00Z","relpermalink":"/blog/seminar_detection_dogs/","section":"blog","summary":"As one of our always excellent department seminar, we had the chance to welcome Heath Smith from University of Washington, Center for Conservation Biology. Heath, together with his dog Chester with whom I became friend during the talk, is working with detection dogs on all sorts of important questions and organisms in conservation biology.","tags":["seminar"],"title":"H. Smith's talk on detection dogs","type":"blog"},{"authors":null,"categories":null,"content":"I acted as the \u0026ldquo;president\u0026rdquo; on the PhD committee of Sophie Monsarrat who, with her supervisor Ana Rodrigues, kindly invited me. It was about the reconstruction of marine mammal’s historical distribution and abundance using historical data. Sophie did an amazing job during her PhD, have a look to her recent paper on the topic. Her presentation was just awesome, I wish I could give such vivid and fascinating talks, irrespective of the topic. Otherwise, her work reminded me of my current work on monk seals with Alex Karamanlidis from the MoM NGO and the talk I\u0026rsquo;m supposed to give at the ICCB2015 on the analysis of citizen science data. More soon.\n","date":1431000000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1431000000,"objectID":"f59e6bdad5bcde0d11fe7343347c640b","permalink":"https://oliviergimenez.github.io/blog/phd_defense/","publishdate":"2015-05-07T12:00:00Z","relpermalink":"/blog/phd_defense/","section":"blog","summary":"I acted as the \u0026ldquo;president\u0026rdquo; on the PhD committee of Sophie Monsarrat who, with her supervisor Ana Rodrigues, kindly invited me. It was about the reconstruction of marine mammal’s historical distribution and abundance using historical data.","tags":["PhD defense"],"title":"PhD defense on historical ecology","type":"blog"},{"authors":null,"categories":null,"content":"Great news! Gilles Maurer will soon start his PhD on the links between wild and captive populations of Asian elephants using an interdisciplinary approach. This is a joint venture with the Beauval zoo and B. Mulot through the CIFRE framework that funds applied PhDs with non-academic partners.\n","date":1429790400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1429790400,"objectID":"ac08c0e8e45171a96f882f40e37494c5","permalink":"https://oliviergimenez.github.io/blog/phd_grant/","publishdate":"2015-04-23T12:00:00Z","relpermalink":"/blog/phd_grant/","section":"blog","summary":"Great news! Gilles Maurer will soon start his PhD on the links between wild and captive populations of Asian elephants using an interdisciplinary approach. This is a joint venture with the Beauval zoo and B.","tags":["PhD grant","socio-ecological approach"],"title":"PhD project on interactions between wild and captive Asian elephant populations","type":"blog"},{"authors":null,"categories":null,"content":"Paméla Lagrange successfully defended her PhD co-supervised with M. Bélisle from the University of Sherbrooke on the \u0026lsquo;Drivers of survival and breeding dispersal using a capture-recapture framework in Tree Swallows - Québec\u0026rsquo;. The members of the committee (M. Festa-Bianchet, G. Gauthier, C. Barbraud, E. Cam) were unanimously amazed by both the ecological and methodological contributions of Paméla to the field. Congrats Pam!\n","date":1428580800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1428580800,"objectID":"0713f224116855a6ef2cbec7c9ec2541","permalink":"https://oliviergimenez.github.io/blog/phd_defense_pam/","publishdate":"2015-04-09T12:00:00Z","relpermalink":"/blog/phd_defense_pam/","section":"blog","summary":"Paméla Lagrange successfully defended her PhD co-supervised with M. Bélisle from the University of Sherbrooke on the \u0026lsquo;Drivers of survival and breeding dispersal using a capture-recapture framework in Tree Swallows - Québec\u0026rsquo;.","tags":["PhD defense"],"title":"Pamela Lagrange's PhD defense","type":"blog"},{"authors":null,"categories":null,"content":"I had the visit of Nina Santostasi from Italy to work for a few days on the estimation of striped dolphin abundance in the gulf of Corinth (in Greece). Nina is working with Giovanni Bearzi and Silvia Bonizzoni from the NGO Dolphin Biology and Conservation. Hopefully our collaboration will provide elements for the conservation of the species.\n","date":1428408000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1428408000,"objectID":"e27aa09c10dab36749db82fd54bb3c17","permalink":"https://oliviergimenez.github.io/blog/nina_visit/","publishdate":"2015-04-07T12:00:00Z","relpermalink":"/blog/nina_visit/","section":"blog","summary":"I had the visit of Nina Santostasi from Italy to work for a few days on the estimation of striped dolphin abundance in the gulf of Corinth (in Greece). Nina is working with Giovanni Bearzi and Silvia Bonizzoni from the NGO Dolphin Biology and Conservation.","tags":["visitor","abundance"],"title":"Nina Santostasi's visit","type":"blog"},{"authors":null,"categories":null,"content":"As part of the GdR EcoStat and our evolutionary demography group, we had a stimulating workshop on the quantification of individual heterogeneity in survival of wild populations using an approach recently developed by Hal Caswell. Hal attended the workshop with Lotte de Vries her new PhD student who will be leading the study.\n","date":1427976000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1427976000,"objectID":"e9182b7d2be29d2060a549b86b0f450e","permalink":"https://oliviergimenez.github.io/blog/gdr_demography/","publishdate":"2015-04-02T12:00:00Z","relpermalink":"/blog/gdr_demography/","section":"blog","summary":"As part of the GdR EcoStat and our evolutionary demography group, we had a stimulating workshop on the quantification of individual heterogeneity in survival of wild populations using an approach recently developed by Hal Caswell.","tags":["demography","working group","individual heterogeneity"],"title":"Working group in demography","type":"blog"},{"authors":null,"categories":null,"content":"We organised the first conference of the GdR EcoStat. We spent two lovely days in Lyon talking about ecological statistics with the hundred people who attended. The talks are available here. If you\u0026rsquo;d like to know more about our activities in the GdR, check out our website.\n","date":1426161600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1426161600,"objectID":"d3fba201c1b9411a52c8138f325ea53d","permalink":"https://oliviergimenez.github.io/blog/gdr_ecostat_conf/","publishdate":"2015-03-12T12:00:00Z","relpermalink":"/blog/gdr_ecostat_conf/","section":"blog","summary":"We organised the first conference of the GdR EcoStat. We spent two lovely days in Lyon talking about ecological statistics with the hundred people who attended. The talks are available here.","tags":["conference","statistical ecology"],"title":"GDR EcoStat first conference","type":"blog"},{"authors":null,"categories":null,"content":"Laetitia Blanc succesfully defended her PhD on the conservation of lynx in France. Congrats Laeti! She had very nice feedbacks by the members of committee (X. Lambin, E. Marboutin, J. Linnell, M. Schaub and F. Bonadonna). Laetitia will be working for a few months with us on the estimation of lynx abundance in Norway (with J. Linnell) and with ONCFS to finish up a chapter of her thesis.\n","date":1423224000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1423224000,"objectID":"753209f0566e9e933dddd35e8ea5ea8a","permalink":"https://oliviergimenez.github.io/blog/laeti_phd_defense/","publishdate":"2015-02-06T12:00:00Z","relpermalink":"/blog/laeti_phd_defense/","section":"blog","summary":"Laetitia Blanc succesfully defended her PhD on the conservation of lynx in France. Congrats Laeti! She had very nice feedbacks by the members of committee (X. Lambin, E. Marboutin, J. Linnell, M.","tags":["PhD defense","lynx"],"title":"Laetitia Blanc defended her PhD","type":"blog"},{"authors":null,"categories":null,"content":"Welcome to our new master students Iago Bonnici and Julie Louvrier and Tamar Lok our new post-doc.\n","date":1422792000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1422792000,"objectID":"e0430b0ae8f4660274b668f82ad0f4ba","permalink":"https://oliviergimenez.github.io/blog/arrivals/","publishdate":"2015-02-01T12:00:00Z","relpermalink":"/blog/arrivals/","section":"blog","summary":"Welcome to our new master students Iago Bonnici and Julie Louvrier and Tamar Lok our new post-doc.","tags":["students"],"title":"Welcome","type":"blog"},{"authors":null,"categories":null,"content":"Our team has recently joined the department Biodiversity \u0026amp; Conservation at CEFE, a department that I have the pleasure to head now. We have moved offices and are now on the second floor. I have an office with a brand new parquet floor and red walls (that I chose and installed/painted myself, can you believe it?!). We had our office-warming party early March. Do not hesitate to come by and visit our new offices!\n","date":1420891200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1420891200,"objectID":"2f7525a72559153ccd559a45270440d4","permalink":"https://oliviergimenez.github.io/blog/move/","publishdate":"2015-01-10T12:00:00Z","relpermalink":"/blog/move/","section":"blog","summary":"Our team has recently joined the department Biodiversity \u0026amp; Conservation at CEFE, a department that I have the pleasure to head now. We have moved offices and are now on the second floor.","tags":["CEFE"],"title":"Moving upstairs","type":"blog"},{"authors":null,"categories":null,"content":"Our meeting report on ISEC2014 has appeared in Biology Letters just before Christmas eve, nice gift! Check it out, it\u0026rsquo;s in open access.\n","date":1419422400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1419422400,"objectID":"d5b00ddb1349017ce3624ccaf25fc87f","permalink":"https://oliviergimenez.github.io/blog/isec_paper/","publishdate":"2014-12-24T12:00:00Z","relpermalink":"/blog/isec_paper/","section":"blog","summary":"Our meeting report on ISEC2014 has appeared in Biology Letters just before Christmas eve, nice gift! Check it out, it\u0026rsquo;s in open access.","tags":["paper","statistical ecology"],"title":"Statistical ecology comes of age","type":"blog"},{"authors":null,"categories":null,"content":"Jim Sedinger, professor at the University of Nevada (Reno), said goodbye to the team after 3 months with us on a sabbatical. It’s been great to have Jim here analyzing data collected as part of the amazing Black Brant monitoring initiated in the 40s. More soon on our joint project on senescence and individual heterogeneity.\n","date":1418731200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1418731200,"objectID":"323243d8d3ad1d196291c594e2b821cd","permalink":"https://oliviergimenez.github.io/blog/jim_sedinger/","publishdate":"2014-12-16T12:00:00Z","relpermalink":"/blog/jim_sedinger/","section":"blog","summary":"Jim Sedinger, professor at the University of Nevada (Reno), said goodbye to the team after 3 months with us on a sabbatical. It’s been great to have Jim here analyzing data collected as part of the amazing Black Brant monitoring initiated in the 40s.","tags":["visitor"],"title":"Jim Sedinger's sabbatical","type":"blog"},{"authors":null,"categories":null,"content":"I acted as president of the examining committee of Sarah Calba who defended her PhD in epistemology of sciences with the predictions in community ecology as a case study. The thesis was supervised by two clever colleagues of mine, Virginie Maris and Vincent Devictor. Sarah did more than well, and the discussions were lively and the debate vivid. As a general comment, I wish we, as ecologists, had more time as part of our training to think of why and how we conduct our scientific activities. Sarah has a nice paper showing that the identification of patterns and processes linking species diversity to functional or phylogenetic diversity depends on the methodological choices we make as analysts. This reminds me of the discussion about \u0026ldquo;Researcher Degrees of Freedom\u0026rdquo; on Gelman’s blog.\n","date":1418731200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1418731200,"objectID":"d90e67a1fa4a648463a1f893b8cfca70","permalink":"https://oliviergimenez.github.io/blog/sarah_calba/","publishdate":"2014-12-16T12:00:00Z","relpermalink":"/blog/sarah_calba/","section":"blog","summary":"I acted as president of the examining committee of Sarah Calba who defended her PhD in epistemology of sciences with the predictions in community ecology as a case study. The thesis was supervised by two clever colleagues of mine, Virginie Maris and Vincent Devictor.","tags":["PhD defense","social sciences"],"title":"Sarah Calba's PhD defense","type":"blog"},{"authors":null,"categories":null,"content":"We held AppliBugs in Montpellier, with the idea to stimulate exchanges and to share information on Bayesian methods and applications ( program here). The day went well despite a rainstorm falling on Montpellier. All talks were great, and I was particularly interested in Adrien Todeschini\u0026rsquo;s presentation of Biips, a software allowing the implementation of state-space models using particle filtering and a R/BUGS-like syntax.\n","date":1417176000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1417176000,"objectID":"eb4bbb495775e5fe7031508928da4a6e","permalink":"https://oliviergimenez.github.io/blog/applibugs_montpellier/","publishdate":"2014-11-28T12:00:00Z","relpermalink":"/blog/applibugs_montpellier/","section":"blog","summary":"We held AppliBugs in Montpellier, with the idea to stimulate exchanges and to share information on Bayesian methods and applications ( program here). The day went well despite a rainstorm falling on Montpellier.","tags":["applibugs","bayes","conference"],"title":"AppliBugs and applications of Bayesian theory","type":"blog"},{"authors":null,"categories":null,"content":"We had a lovely 2-day workshop in Lyon on individual heterogeneity in demography as part of our research group on Statistical Ecology. I gave a short talk providing an overview of heterogeneous capture-recapture models. We agreed on meeting again early April 2015 to carry out a comparative analysis of individual heterogeneity using the datasets shared by the group and the method developed by Hal Caswell (check it out here).\n","date":1416484800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1416484800,"objectID":"f5e07dec086d05e914d29ff6691be905","permalink":"https://oliviergimenez.github.io/blog/idh_meeting/","publishdate":"2014-11-20T12:00:00Z","relpermalink":"/blog/idh_meeting/","section":"blog","summary":"We had a lovely 2-day workshop in Lyon on individual heterogeneity in demography as part of our research group on Statistical Ecology. I gave a short talk providing an overview of heterogeneous capture-recapture models.","tags":["workshop","individual heterogeneity"],"title":"Workshop on individual heterogeneity","type":"blog"},{"authors":null,"categories":null,"content":"After a month off, I’m back to work starting with a 2-day workshop on citizen sciences I’ve co-organised with Romain Julliard and Pascal Monestiez ( program) as part of our group on Statistical Ecology and on citizen sciences CiSStats. The discussions were stimulating with a nice balance between talks given by data providers on the first day and modelers on the second day. I gave a talk based on Anne-Sophie’s work on mapping brown bear distribution in Greece using opportunistic data. Slides below:\nCheck out below Anne-Sophie\u0026rsquo;s talk she gave at the 2014 International Conference on Bear Research and Management in Thessaloniki (Greece) and at the joint BES-SFE Ecology conference in Lille.\n","date":1413460800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1413460800,"objectID":"1e6664c7af342984f035a720b77e6d19","permalink":"https://oliviergimenez.github.io/blog/cs_meeting/","publishdate":"2014-10-16T12:00:00Z","relpermalink":"/blog/cs_meeting/","section":"blog","summary":"After a month off, I’m back to work starting with a 2-day workshop on citizen sciences I’ve co-organised with Romain Julliard and Pascal Monestiez ( program) as part of our group on Statistical Ecology and on citizen sciences CiSStats.","tags":["workshop","citizen science"],"title":"Workshop on citizen science and statistics","type":"blog"},{"authors":null,"categories":null,"content":"I attended Daniel Turek\u0026rsquo;s talk at ISEC about NIMBLE a neat alternative to WinBUGS and JAGS. It is developed by Perry de Valpine\u0026rsquo;s group at Berkeley and \u0026lsquo;lets you use BUGS models natively in R, program functions that use them, and compile everything via C++ for faster computing\u0026rsquo;. I played around with NIMBLE a bit, here is an example of fitting a classic capture-recapture model to simulated data - thanks to Perry and Daniel for their help! NIMBLE seems a lot faster than its competitors, and much more flexible. I\u0026rsquo;ll continue my investigations with more complex models - stay tuned.\n","date":1404993600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1404993600,"objectID":"0869115bd7d03194d6a945b2759b7623","permalink":"https://oliviergimenez.github.io/blog/nimble/","publishdate":"2014-07-10T12:00:00Z","relpermalink":"/blog/nimble/","section":"blog","summary":"I attended Daniel Turek\u0026rsquo;s talk at ISEC about NIMBLE a neat alternative to WinBUGS and JAGS. It is developed by Perry de Valpine\u0026rsquo;s group at Berkeley and \u0026lsquo;lets you use BUGS models natively in R, program functions that use them, and compile everything via C++ for faster computing\u0026rsquo;.","tags":["isec","R","conference"],"title":"New kids on the Bayesian block","type":"blog"},{"authors":null,"categories":null,"content":"ISEC 2014 is now closed. Organising this conference has been a fantastic adventure; hopefully the participants liked it.\n","date":1404561600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1404561600,"objectID":"30ad5fd73edef01a3cb2561964b0f1f1","permalink":"https://oliviergimenez.github.io/blog/isec2014_end/","publishdate":"2014-07-05T12:00:00Z","relpermalink":"/blog/isec2014_end/","section":"blog","summary":"ISEC 2014 is now closed. Organising this conference has been a fantastic adventure; hopefully the participants liked it.","tags":["isec","conference"],"title":"ISEC 2014: the end","type":"blog"},{"authors":null,"categories":null,"content":"I gave a talk at ISEC 2014 on fitting occupancy models with hidden-Markov models using E-SURGE:\n","date":1404561600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1404561600,"objectID":"fd3bf509d9243080e128d52f4f50c0ad","permalink":"https://oliviergimenez.github.io/blog/isec2014_talks/","publishdate":"2014-07-05T12:00:00Z","relpermalink":"/blog/isec2014_talks/","section":"blog","summary":"I gave a talk at ISEC 2014 on fitting occupancy models with hidden-Markov models using E-SURGE:","tags":["isec","conference"],"title":"We're presenting our work at ISEC 2014","type":"blog"},{"authors":null,"categories":null,"content":"Ben Rashford and Thomas Foulke from the University of Wyoming visited us. They gave a great talk on \u0026ldquo;Wolf Re-Introduction in the Northern Rockies, USA: From Population and Predation Trends to Policy and Economics\u0026rdquo;.\n","date":1403697600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1403697600,"objectID":"9f0e47bb3be21f80639f93c9acd2870e","permalink":"https://oliviergimenez.github.io/blog/economy_wolf/","publishdate":"2014-06-25T12:00:00Z","relpermalink":"/blog/economy_wolf/","section":"blog","summary":"Ben Rashford and Thomas Foulke from the University of Wyoming visited us. They gave a great talk on \u0026ldquo;Wolf Re-Introduction in the Northern Rockies, USA: From Population and Predation Trends to Policy and Economics\u0026rdquo;.","tags":["visitor","talk"],"title":"Economy and large carnivores","type":"blog"},{"authors":null,"categories":null,"content":"Rachel Guénon successfully defended her Master internship on the Dalmatian pelican and life-history tradeoffs (from Greece!). This is joint work with Alain Crivelli from Tour du Valat.\n","date":1403352000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1403352000,"objectID":"dc95c25df80bf6f31192a1e86589dbcf","permalink":"https://oliviergimenez.github.io/blog/pelican_rachel/","publishdate":"2014-06-21T12:00:00Z","relpermalink":"/blog/pelican_rachel/","section":"blog","summary":"Rachel Guénon successfully defended her Master internship on the Dalmatian pelican and life-history tradeoffs (from Greece!). This is joint work with Alain Crivelli from Tour du Valat.","tags":["master","defense"],"title":"Pelican and life-history trade-offs","type":"blog"},{"authors":null,"categories":null,"content":"Sending questionnaires for Blaise\u0026rsquo;s PhD to study the local perception of the presence of brown bears in the Pyrénées. First picture, the 3000 questionnaires to be sent; second picture, those questionnaires once packed in boxes; third picture, the nice guy from the mail company La Poste comes and picks them up. Now the question is how many of them we will we get back?\u0026hellip;\n","date":1403265600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1403265600,"objectID":"656a87be874c55a606cb08cd9065131b","permalink":"https://oliviergimenez.github.io/blog/blaise_quest/","publishdate":"2014-06-20T12:00:00Z","relpermalink":"/blog/blaise_quest/","section":"blog","summary":"Sending questionnaires for Blaise\u0026rsquo;s PhD to study the local perception of the presence of brown bears in the Pyrénées. First picture, the 3000 questionnaires to be sent; second picture, those questionnaires once packed in boxes; third picture, the nice guy from the mail company La Poste comes and picks them up.","tags":["questionnaire","social sciences","brown bear","attitude"],"title":"Studying attitudes towards large carnivores","type":"blog"},{"authors":null,"categories":null,"content":"The team is leading the organisation of ISEC2014. We\u0026rsquo;re getting ready by, among many other things, preparing the delegates\u0026rsquo; packs. Of course, these efforts cannot go without some relaxing time.\n","date":1401624000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1401624000,"objectID":"961fbe75d08146fe987ac1666ffdd021","permalink":"https://oliviergimenez.github.io/blog/isec2014_prep/","publishdate":"2014-06-01T12:00:00Z","relpermalink":"/blog/isec2014_prep/","section":"blog","summary":"The team is leading the organisation of ISEC2014. We\u0026rsquo;re getting ready by, among many other things, preparing the delegates\u0026rsquo; packs. Of course, these efforts cannot go without some relaxing time.","tags":["isec","conference"],"title":"We're organising ISEC 2014","type":"blog"},{"authors":null,"categories":null,"content":"Relaxing after finishing up writing a big grant proposal. Tedeschi and Truck concert at the mytic Olympia Paris with Jonny Lang as the opening act.\n","date":1398686400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1398686400,"objectID":"b606bacb0d3a19c4f4962fc166e0afaf","permalink":"https://oliviergimenez.github.io/blog/relax_concert/","publishdate":"2014-04-28T12:00:00Z","relpermalink":"/blog/relax_concert/","section":"blog","summary":"Relaxing after finishing up writing a big grant proposal. Tedeschi and Truck concert at the mytic Olympia Paris with Jonny Lang as the opening act.","tags":["relaxing","music"],"title":"Post grant submission relaxing","type":"blog"},{"authors":null,"categories":null,"content":"Dave and Lise are going back home to the US after a 6-month sabbatical with us. It\u0026rsquo;s been awesome to have you here!\n","date":1397131200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1397131200,"objectID":"95fadecc646087f2dc46aefb0e7fc3f0","permalink":"https://oliviergimenez.github.io/blog/lise_dave/","publishdate":"2014-04-10T12:00:00Z","relpermalink":"/blog/lise_dave/","section":"blog","summary":"Dave and Lise are going back home to the US after a 6-month sabbatical with us. It\u0026rsquo;s been awesome to have you here!","tags":["visitor"],"title":"Bye bye Lise and Dave","type":"blog"},{"authors":null,"categories":null,"content":"The team run two 1-week workshops on capture-recapture models and matrix models. We\u0026rsquo;re grateful to all the attendees who came, with fascinating research topics and stimulating questions. No need to say that your presence means a lot to us! We hope to see again many people next year! Warm and special thanks to Dave Koons, Lise Aubry and Pamela Lagrange!\n","date":1395316800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1395316800,"objectID":"aaf78930310de111ec9f0518f9a2fe8b","permalink":"https://oliviergimenez.github.io/blog/workshop_cmr_2014/","publishdate":"2014-03-20T12:00:00Z","relpermalink":"/blog/workshop_cmr_2014/","section":"blog","summary":"The team run two 1-week workshops on capture-recapture models and matrix models. We\u0026rsquo;re grateful to all the attendees who came, with fascinating research topics and stimulating questions. No need to say that your presence means a lot to us!","tags":["workshop","capture-recapture"],"title":"Capture-recapture workshop","type":"blog"},{"authors":null,"categories":null,"content":"The French research group in Statistical Ecology (GDR EcoStat) is born. Funded by CNRS, the objective is to structure the research in statistical ecology at a national (French) level. Check out our website, and sign up for our mailing list!\n","date":1388577600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1388577600,"objectID":"55b97448b70522e9e90cdd3c31736451","permalink":"https://oliviergimenez.github.io/blog/gdr_ecostat_is_born/","publishdate":"2014-01-01T12:00:00Z","relpermalink":"/blog/gdr_ecostat_is_born/","section":"blog","summary":"The French research group in Statistical Ecology (GDR EcoStat) is born. Funded by CNRS, the objective is to structure the research in statistical ecology at a national (French) level. Check out our website, and sign up for our mailing list!","tags":["statistical ecology"],"title":"The GDR is statistical ecology is born!","type":"blog"},{"authors":null,"categories":null,"content":"Below are the slides of my plenary talk at the 2013 EURING meeting on individual variability in capture-recapture models. I couldn’t attend, but the organizers were kind enough to run my presentation to the audience. I had to be a bit imaginative for the attendees not to fall asleep\u0026hellip;\n","date":1367755200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1367755200,"objectID":"3b97d937868e18c046753f2866b2d155","permalink":"https://oliviergimenez.github.io/blog/euring_2013/","publishdate":"2013-05-05T12:00:00Z","relpermalink":"/blog/euring_2013/","section":"blog","summary":"Below are the slides of my plenary talk at the 2013 EURING meeting on individual variability in capture-recapture models. I couldn’t attend, but the organizers were kind enough to run my presentation to the audience.","tags":["statistical ecology","individual heterogeneity","conference"],"title":"Individual heterogeneity and capture-recapture at EURING 2013","type":"blog"},{"authors":null,"categories":null,"content":"Here are the slides of my interview for a CNRS senior scientist position (DR2). Note that CNRS offers interdisciplinary positions, which is sadly not true for all of our research institutes.\n","date":1334923200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1334923200,"objectID":"91577d3b3c1cd18a77502b695e36f28c","permalink":"https://oliviergimenez.github.io/blog/cnrs_interview/","publishdate":"2012-04-20T12:00:00Z","relpermalink":"/blog/cnrs_interview/","section":"blog","summary":"Here are the slides of my interview for a CNRS senior scientist position (DR2). Note that CNRS offers interdisciplinary positions, which is sadly not true for all of our research institutes.","tags":["cnrs","interview"],"title":"My CNRS interview for a senior scientist position","type":"blog"},{"authors":null,"categories":null,"content":"Here are the slides of my HDR defense. This is a necessary step in France (and in other countries) to be authorized to supervise (officially) PhD students.\n","date":1269777600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1269777600,"objectID":"d47f53e69ec744817d70a56d7fa14d6f","permalink":"https://oliviergimenez.github.io/blog/hdr_defense/","publishdate":"2010-03-28T12:00:00Z","relpermalink":"/blog/hdr_defense/","section":"blog","summary":"Here are the slides of my HDR defense. This is a necessary step in France (and in other countries) to be authorized to supervise (officially) PhD students.","tags":["hdr","defense"],"title":"My HDR defense","type":"blog"},{"authors":null,"categories":null,"content":"Below I provide pieces of code I have written over the years and shared through GitHub on reproducible science, capture-recapture, occupancy and epidemiological models, spatial stuff, social networks, deep learning, textual analyses, the tidyverse and some random stuff.\nThis code is provided as is without warranty of any kind. If you find a bug or anything, feel free to get in touch or fill an issue on GitHub.\nYou may find other pieces of code in papers.\nLast update: August 2021.\nReproducible science Introduction to Git and GitHub with RStudio [ slides] [ material]\nWriting dynamic and reproducible documents with R Markdown [ slides] [ material] [ practical] [ interactive dashboard] [ website]\nManipulate and visualise data in the tidyverse [ slides] [ material] [ tips]\nFocus on GIS and spatial data with sf package [ slides] [ material]\nCapture-recapture Quantifying bias in mortality estimates when ignoring imperfect detection \u0026ndash; R, Shiny\nCapture-recapture models as HMMs \u0026ndash; R, Jags\nSpatial capture-recapture models for closed and open populations \u0026ndash; R, Nimble\nJolly-Seber capture-recapture models \u0026ndash; R, Jags\nMultievent capture-recapture models \u0026ndash; R, Jags, Nimble\nMultievent capture-recapture models with Rcpp \u0026ndash; R, Rcpp\nMultievent SIR models and capture-recapture data \u0026ndash; R, TMB\nLocal minima and multistate capture-recapture models; see also there for mixture of recaptures and recoveries \u0026ndash; R, Jags, ADMB, Stan, unmarked\nAnimal social networks and capture-recapture data \u0026ndash; R, Jags\nIndividual fitness in the wild and capture-recapture data \u0026ndash; R, Jags, E-SURGE\nGompertz capture-recapture model with a Gamma frailty \u0026ndash; R, Jags\nMany correlated covariates in capture-recapture models \u0026ndash; R, RMark\nStructural equation capture-recapture models \u0026ndash; R, Jags\nAbundance estimation using capture-recapture models \u0026ndash; R, RMark\nBand-recovery models with time random effects in a frequentist framework \u0026ndash; R\nBayesian implementation of capture-recapture models with robust design \u0026ndash; R, Jags\nPredator-prey integrated community models \u0026ndash; R, Jags\nOccupancy Dynamic analysis of brown bear habitat use in the Pyrenees \u0026ndash; R, unmarked\nDynamic occupancy models in TMB and ADMB \u0026ndash; R, TMB, ADMB\nMultistate occupancy model with uncertainty \u0026ndash; R, Jags\nHMM formulation of a multispecies dynamic occupancy models \u0026ndash; R\nSimulation and fit of a 2-species occupancy model à la Rota et al. (2016) here with unmarked and there with Jags and Nimble\nBias in occupancy estimate for a static occupancy model, with an interactive dashboard \u0026ndash; R and unmarked\nSpatial Animated map of wolf presence in France \u0026ndash; R\nDistribution map from an occupancy model, with Eurasian or Balkan lynx \u0026ndash; R, unmarked\nIntroduction to GIS and mapping in R using the sf package with slides \u0026ndash; R\nEpidemiological models and ODEs Dynamic models based on ODEs here and there \u0026ndash; Nimble\nExplore strategies of social distancing on epidemic w/ SIR model \u0026ndash; R\nZombie apocalypse and a Bayesian approach for fitting an epidemiological model to data \u0026ndash; R, OpenBUGS\nSocial networks Network analysis of my Twitter account \u0026ndash; R\nNetwork analysis of Star Wars The Attack of the Clones \u0026ndash; R\nAnimal social networks from capture-recapture data\nAnalysis of Bottlenose dolphin social network \u0026ndash; R\nScientific research is all about networking \u0026ndash; R\nDeep learning and species identification Piégeage photo et pipeline pour l\u0026rsquo;identification d\u0026rsquo;espèces via deep learning \u0026ndash; Python, R Bibliometric and textual analyses Content analysis of French articles in newspapers about wolf \u0026ndash; R\nBibliometric and textual analyses of the last decade of research in capture-recapture \u0026ndash; R\nQuick and dirty analyses of ISEC 2018 tweets \u0026ndash; R\nText mining of the ISEC 2020 abstracts \u0026ndash; R\nTidyverse Introduction to the tidyverse \u0026ndash; R\nSome notes I have taken on David Robinson\u0026rsquo;s screencasts, with R tidyverse tips and tricks \u0026ndash; R\nAnalyses of wolf depredation in France \u0026ndash; R\nVisualizing trends in French deaths \u0026ndash; R\nRandom stuff Decoding the hidden message in Perseverance parachute with R \u0026ndash; R\nVisualising the Metropolis algorithm on a simple 1D example \u0026ndash; R\nExport, clean up and tabulate questions and answers from a Slack work space with several channels into HTML tables \u0026ndash; R and json\nIllustration of the delta-method \u0026ndash; R\nCreating a hex sticker for the R2ucare package \u0026ndash; R\nSimulate data with Jags \u0026ndash; R, Jags\nFirst steps in Python \u0026ndash; Python\nBayesian Multivariate Autoregressive State-Space models - MARSS \u0026ndash; R, Jags\n","date":-62135596800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":-62135596800,"objectID":"fcef553317377ec10e2c25c685386d28","permalink":"https://oliviergimenez.github.io/coding/","publishdate":"0001-01-01T00:00:00Z","relpermalink":"/coding/","section":"","summary":"Below I provide pieces of code I have written over the years and shared through GitHub on reproducible science, capture-recapture, occupancy and epidemiological models, spatial stuff, social networks, deep learning, textual analyses, the tidyverse and some random stuff.","tags":null,"title":"Random pieces of code","type":"page"}] \ No newline at end of file +[{"authors":["admin"],"categories":null,"content":"My name is Olivier Gimenez and I am a CNRS senior scientist at CEFE working at the interface of animal ecology, statistical modeling and social sciences in the Human-Animal Interactions group. I studied maths and statistics in particular, finishing my PhD in 2003. I obtained my Habilitation in ecology and evolution, and more recently embarked in studying sociology.\nWith students and colleagues, we focus on questions in ecology and conservation biology dealing with the impact of climate change and human activities on population dynamics and species distribution and the management of large mammals. I\u0026rsquo;m also interested in human-wildlife interactions and how interdisciplinary approaches may contribute to the coexistence of humans and animals.\nIf you\u0026rsquo;d like to work with us, feel free to contact me, and check out this paper which gives the spirit of how we enjoy doing science. For news and chat, you may find me on Twitter, Mastodon and GitHub 😄. When time allows, I write R posts which are also published on rweekly and r-bloggers.\n","date":-62135596800,"expirydate":-62135596800,"kind":"term","lang":"en","lastmod":-62135596800,"objectID":"2525497d367e79493fd32b198b28f040","permalink":"https://oliviergimenez.github.io/authors/admin/","publishdate":"0001-01-01T00:00:00Z","relpermalink":"/authors/admin/","section":"authors","summary":"My name is Olivier Gimenez and I am a CNRS senior scientist at CEFE working at the interface of animal ecology, statistical modeling and social sciences in the Human-Animal Interactions group.","tags":null,"title":"","type":"authors"},{"authors":null,"categories":null,"content":"\nCélian Monchy (PhD 2022-2025) is working on the propagation of uncertainties related to the automatic processing of sensor data in different models in ecology. He is co-supervised by Marie-Pierre Etienne.\nLéa Pautrel (PhD 2022-2025) is investigating the use of deep learning coupled with continuous-time models to exploit sensor data, in particular cam traps. She is co-supervised by Marie-Pierre Etienne and Sylvain Moulherat from TerrOïko.\nMaëlis Kervellec (PhD 2021-2024) is developing statistical models for the analysis of camera trap data with applications in large carnivores population ecology. This is joint work with colleagues from the French Office of Biodiversity\nMarwan Naciri (PhD 2021-2024) is studying the demography of polar bears. He is co-supervised by Sarah Cubaynes and Jon Aars.\nOksana Grente (PhD 2018-2021, post-doc 2023-2024) works at evaluating the impact of lethal control on the French wolf population, with a special interest in social structure and livestock depredation. This is joint work with Sarah Bauduin, Christophe Duchamp and Nolwenn Hoguet-Drouet from French Office of Biodiversity.\nSimon Lacombe (PhD 2023-2026) is a quantitative ecologists who\u0026rsquo;s determining the drivers of European otter recolonization in France using non-invasive sampling and statistical ecology. He is co-supervised by Sébastien Devillard.\nJavi Fernández-López (post-doc 2022-2023) is interested in conservation biology and phylogeography. He is modelling the distribution of mammals in Europe by integrating various sources of data. Check out his website https://allthiswasfield.blogspot.com.\nMónia Nakamura Mercier Real (PhD 2019-2024) is investigating the impacts of anthropogenic factors on large carnivores persistence in human-dominated landscapes. She uses wolf in Spain and Portugal as a model species to address how different humanization levels can influence habitat selection and land use, and ultimately affect its demography, status and viability. The thesis is co-supervised with José Vicente López-Bao and Pedro Monterroso.\nBéatrice Chataigner (EPHE diploma 2020-2022) is working on locally-based monitoring programs in Kenya. Her research project aims at exploring the performances of current ecological monitoring program of Kenyan conservanciesin addressing their needs in terms of adaptive management at local level as well as ecosystem level. Béatrice works for IUCN in the Program on African Protected Areas and Conservation. Her work is jointly supervised by Nicholas Pilfold and Aurélien Besnard.\nPatrícia Rodrigues (post-doc 2021-2023) is interested in the social-ecological dynamics underpinning environmental problems in Africa. Her current project aims to characterize landscapes within Africa’s Great Green Wall in terms of their social-ecological properties and restoration initiatives. More here.\n","date":1708646400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1708646400,"objectID":"ad001a11940e7f03984cedd862638782","permalink":"https://oliviergimenez.github.io/people/current/","publishdate":"2024-02-23T00:00:00Z","relpermalink":"/people/current/","section":"people","summary":"Célian Monchy (PhD 2022-2025) is working on the propagation of uncertainties related to the automatic processing of sensor data in different models in ecology. He is co-supervised by Marie-Pierre Etienne.","tags":null,"title":"Current lab members","type":"people"},{"authors":null,"categories":null,"content":"\nA list of my former Master students is available here.\nMatthew Silk (post-doc 2022-2023) is a behavioural and disease ecologist researching the links between population dynamics and social structure by developing new social network and demographic models. Matt was a MCSA research fellow. He is now a research fellow in Edinburgh.\nLouis Barbu (intern 2022) was a second year student of BTSA GPN. Louis conducted a study on the functionality of ecological corridors through the European otter in the Hérault department.\nJulie Louvrier (post-doc 2022) is an ecologist with strong quantitative skills. She worked on the interactions between large carnivores using occupancy models.\nMarine Desprez (post-doc 2022) worked on machine learning methods for classification problems in ecology. This was joint work with Vincent Miele.\nAlix Cosquer (post-doc 2019-2022) is an environmental psychologist. She studied how to favor connections between nature and children using educative marine areas, and worked on children\u0026rsquo;s relationships with nature in schools.\nMaud Quéroué (PhD 2018-2021) developed multispecies demographic models to study predator-prey interactions in seabirds and competition/coexistence in passerines. This was joint work with Christophe Barbraud, Pierre-Yves Henry and Frédéric Barraquand. Maud is now a post-doc with Aurélien Besnard.\nValentin Lauret (PhD 2018-2021) developed integrated models for monitoring of bottlenose dolphins in the network of marine protected areas in the French Mediterranean Sea. This was joint work with Hélène Labach from the MIRACETI NGO. Valentin is now a post-doc with Aurélien Besnard.\nHélène Labach (PhD 2017-2021) from GIS3M was working on the conservation of bottlenose dolphins in the North-Western Mediterranean Sea. Her work relied on the data collected through the GDEGeM project. She continues her work as the head of the MIRACETI NGO.\nNina Santostasi (PhD 2016-2019) developed capture-recapture and demographic models for studying hybridization in the wild. As case studies, she considered dolphins in the Gulf of Corinth in Greece (with Giovanni Bearzi and Silvia Bonizzoni from Dolphin Biology and Conservation) and wolves in Italy (with Paolo Ciucci from University of Rome - La Sapienza). Nina is now a post-doc with Paolo Ciucci working on estimating population abundance of brown bears.\nSarah Bauduin (post-doc 2018-2019) built spatially-explicit agent-based models to assess the viability of Eurasian lynx in France, and more broadly to study multispecies demography. Sarah is now a permanent researcher with the French Office of Biodiversity.\nGilles Maurer (PhD 2015-2018, post-doc 2018-2020) studied the interactions between wild and semi-captive pools of Asian elephants with an interdisciplinary framework (demography, economy, anthropology, population genetics). This was joint work with Baptiste Mulot at Beauval Nature, Finn Kjellberg and Nicolas Lescureux from CEFE. Gilles is now a permanent researcher at Beauval Nature, hosted at CEFE.\nLorena Mansilla (PhD 2016-2019) developed capture-recapture and demographic models for social species with wolves, dolphins and elephant seals as case studies. She used spatially-explicit capture-recapture models, social networks, cluster analyses and integrated population models (her thesis was co-supervised with Roger Pradel).\nVincenzo Gervasi (post-doc 2017-2019) used an interdisciplinary approach to understand coexistence patterns and build trans-boundary management tools for coexistence between large carnivores and human activities in Europe. This was joint work with John Linnell from NINA (Norway) and Luigi Boitani from University of Rome - La Sapienza. Vincenzo is now a post-doc with ISPRA back in Italy working on the African swine fever epidemics in wild boar.\nJulie Louvrier (PhD 2015-2018) evaluated and developed occupancy models for assessing the distribution of large carnivores (wolf, lynx) in Europe. This was joint work with Christophe Duchamp, Nolwenn Hoguet-Drouet and Eric Marboutin from ONCFS. Julie is now a post-doc at the Leibniz Institute for Zoo and Wildlife Research.\nLoreleï Guéry (post-doc 2016-2018) studied the dynamics of lynx in Switzerland using spatially-explicit capture-recapture models. This was joint work with Fridolin Zimmermann from KORA. Loreleï then did a post-doc in the MARBEC lab, working with Daniel Gaertner from IRD on the population dynamics of tunas. She is now a permanent researcher at CIRAD.\nBlaise Piédallu (PhD 2013-2016) developed an integrated approach for managing human-wildlife conflicts with the Pyrenean brown bear population as a case study (joint work with Pierre-Yves Quenette from ONCFS). Blaise now works as a biology teacher.\nLaetitia Blanc (PhD 2012-2016, ATER 2016) Worked on the conservation of lynx in France using spatially-explicit individual-based models (joint work with Stephanie Kramer-Schadt from IZW and Eric Marboutin from ONCFS). Laetitia now works as a biology teacher.\nPaméla Lagrange (PhD 2011-2015) studied the drivers of survival and dispersal in Tree Swallows using capture-recapture data (joint work with Marc Bélisle at Sherbrooke). Pam is now working as a project manager at Réserves Naturelles de France.\nTamar Lok (post-doc 2015-2017) developed population models to combine colour-banding, tracking and count data to assess within- and between-individual variation in different aspects of migratory behaviour of the Eurasian Spoonbill and the Red Knot. Tamar is back to the Netherlands where she is a researcher.\nJulien Papaïx (post-doc 2014) assessed the impact of global change on the population dynamics of Atlantic salmons (joint work with Etienne Prévost). Julien is now a permanent researcher at INRA.\nFitsum Abadi (post-doc 2012) developed neat integrated population models. After a post-doc and occupying a lecturer position there, Fitsum is now assistant professor in the US, New Mexico State University.\nSabrina Servanty (post-doc 2008-2009) studied the dynamic of exploited populations with wild boar as a case study. Sabrina then did a post-doc in the US, and a post-doc in St-Pée/Nivelle on the Côte Basque.\nElena Papadatou (post-doc 2009) worked at quantifying environmental variation and its impact on population dynamics of bats. She worked as an environmental expert in Greece for some time, then moved to the UK to take a project manager position at Arup.\nMarlène Gamelon (PhD 2011-2013) studied environmental variation and reproduction tactics in wild boars (joint work with Jean-Michel Gaillard and Eric Baubet). She did a post-doc in Norway. Marlène is now a permanent researcher at CNRS.\nLucile Marescot (PhD 2010-2012, post-doc 2017-2019) analyzed the recolonization of wolves in France (joint work with Eric Marboutin and Guillaume Chapron). She did a post-doc in the US and another post-doc in Germany. Then she developed multi-species occupancy models for large mammals. This was a collaboration with Christophe Duchamp and Nolwenn Hoguet-Drouet from ONCFS (France) and Arnaud Lyet from WWF-US. After working as a freelance data scientist in conservation biology, Lucile got a permanent position as a researcher at CIRAD.\nSarah Cubaynes (PhD 2009-2011, post-doc 2017) developed capture-recapture mixed models (joint work with Christian Lavergne). After a post-doc both in the UK at Oxford and in the US working in the Yellowstone, she did another post-doc with us. Sarah is now assistant professor.\nMathieu Buoro (PhD 2010-2012) worked in evolutionary ecology using mark-recapture data in Atlantic salmons (joint work with Etienne Prévost). Mathieu did a post-doc in the US at Berkeley, then came back as a post-doc in Toulouse. He is now a permanent researcher at INRAE.\nFrançois Guilhaumon (PhD 2008-2010) analysed biodiversity patterns at large scale and worked in conservation biogeography (joint work with David Mouillot). François is now a permanent researcher at IRD.\n","date":1689984000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1689984000,"objectID":"d56f13f3db88f747133e686977ff200c","permalink":"https://oliviergimenez.github.io/people/alumni/","publishdate":"2023-07-22T00:00:00Z","relpermalink":"/people/alumni/","section":"people","summary":"A list of my former Master students is available here.\nMatthew Silk (post-doc 2022-2023) is a behavioural and disease ecologist researching the links between population dynamics and social structure by developing new social network and demographic models.","tags":null,"title":"Former lab members","type":"people"},{"authors":null,"categories":null,"content":" Loutre, Lez, connectivité écologique et médiation scientifique. December 2023. Talk at a meeting between Montpellier city/metropolis and our lab. You can download the slides here and video recording there in French.\nTour d\u0026rsquo;horizon des modèles et de leurs applications. November 2023. Talk a conference co-organised by Cerema, OFB, and centre de ressources TVB et ITTECOP. You can download the slides here in French.\nStatistical ecology, hidden Markov models and the management of large carnivores in Europe. June 28, 2022. Plenary talk for the ISEC2022 conference. You can download the slides and video recording in English.\nDeep learning, false negatives/positives and predator-prey interactions: Peregrinations in R. March 31, 2021. Slides (F = full screen and P = transcript) and video recording in English. All material here. Rencontres R.\nDeep learning, false negatives/positives and predator-prey interactions: Peregrinations in R. March 31, 2021. Slides here (F = full screen and P = transcript) and material there. Video in French. Journées Labex Montpellier.\nLarge carnivores and statistical ecology. Seminar in Lisbon (remotely) Dec 2020. Video and slides.\nFirst steps in Deep Learning - Species identification and camera traps. Speed talk at imaginecology. November 16-17, 2020, Lyon. Video, slides and workflow.\nFlexible spatio-temporal modelling of occupancy using GAMs: A case study with large carnivores. Speed talk at ISEC. 2020, virtual Sydney Australia. Video and slides.\nGrands carnivores et écologie statistique. Séminaire grand public 2020, Rennes France. Vidéo et diapos.\nModélisation spatio-temporelle de l’aire d’occupation : le cas des grands carnivores en France. Annual meeting of the French network in Statistical ecology. 2020, Rennes France. Slides.\nA review of research in capture-recapture: Ecological questions and methods. Wildlife Research and Conservation. 2019, Berlin Germany. Slides, video and material.\nLe projet INTERACT : Interactions entre dauphins et activités humaines sur le littoral méditerranéen français. Rencontres humains - cétacés. 2019, Nice France. Diapos et vidéos.\nInferring animal social networks with imperfect detection. World Marine Mammal Conference. 2019, Barcelona Spain. Slides.\nVariable selection in capture-recapture models using the Lasso. International Statistical Ecology Conference. 2018, St Andrews Scotland. Slides.\nHumains, cétacés et recherche académique. Rencontres sur les Interactions entre Humains et Cétacés. 2018, Marseille France. Diapos et vidéo.\nLe loup dans la presse : approche interdisciplinaire. Séminaire linguistique et écologie. 2018, Montpellier France. Diapos.\nTenter l\u0026rsquo;interdisciplinarité dans le cadre d\u0026rsquo;une thèse sur les conflits humains-ours. Séminaire interdisciplinarité écologie – SHS. 2017, Grenoble France. Diapos et vidéo.\nDealing with many correlated covariates in capture-recapture models. International Statistical Ecology Conference. 2016, Seattle USA. Slides.\nConservation and management of large carnivores in France: A beneficial collaboration between ONCFS practitioners and CNRS researchers. International Conference of Ecological Sciences. 2016, Marseille France. Slides.\nAbundance of bottlenose dolphins along the French Mediterranean coast. International workshop on Bottlenose dolphin conservation and monitoring in the North-Western Mediterranean Sea. 2015, Marseille France. Slides here and there.\nMaking sense of citizen science data: A review of methods. International Congress for Conservation Biology. 2014, Montpellier France. Slides.\nHidden Markov modelling of occupancy data. International Statistical Ecology Conference. 2014, Montpellier France. Slides.\nAll equal, really? Individual variability in capture-recapture models from biological and methodological perspectives. EURING analytical meeting. 2013, Athens USA. Slides.\nEstimating fitness in the wild. International Statistical Ecology Conference. 2012, Oslo Norway. Slides.\nAssessing Inter-population Variation in Survival Using Mark-Recapture Data. Seminar at French National Museum of Natural History. 2011, Paris France. Slides.\nDes modèles de capture-recapture pour l\u0026rsquo;écologie évolutive. Habilitation à Diriger des Recherches. 2010, Montpellier France. Diapos et manuscript.\nCombining longitudinal data and pedigree information to assess heritability of survival. 2009, Pescara Italy. Slides.\nImportance of accounting for detection heterogeneity when estimating abundance: the case of French wolves (Canis lupus). Seminar. 2009, Berkeley USA. Slides.\nIndividual heterogeneity in mortality for incomplete longitudinal data. Workshop in Biodemography. 2009, Stanford USA. Slides.\nPath analysis and structural equation modeling of mark-recapture data. International Statistical Ecology Conference. 2008, St Andrews Scotland. Slides.\nNonparametric estimation of survival using penalized splines: Applications in ecology and evolution. EURING analytical meeting 2007 Dunedin New Zealand, International Biometric Conference 2008 Dublin Ireland, Conference of the French Society for Biometry 2005 Paris France. Slides.\nInvestigating Parameter-Redundancy by Yourself 2003, Radolfzell Germany. Slides.\nEstimation, inference and goodness-of-fit tests for multistate capture-recapture models. PhD defense. 2003, Montpellier France. Slides and manuscript.\n","date":1702252800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1702252800,"objectID":"d3efad569ef55bce750d41b479791de1","permalink":"https://oliviergimenez.github.io/talks/talk/","publishdate":"2023-12-11T00:00:00Z","relpermalink":"/talks/talk/","section":"talks","summary":"Loutre, Lez, connectivité écologique et médiation scientifique. December 2023. Talk at a meeting between Montpellier city/metropolis and our lab. You can download the slides here and video recording there in French.","tags":null,"title":"Talks","type":"talks"},{"authors":null,"categories":null,"content":" Preprints\nMoulherat S., L. Pautrel, G. Debat, M.-P. Etienne, L. Gendron, N. Hautière, J.-P. Tarel, G. Testud, O. Gimenez (2023). Biodiversity Monitoring with Intelligent Sensors: An Integrated Pipeline for Mitigating Animal-Vehicle Collisions. Arpha Preprint.\nBauduin S., Germain E., Zimmermann F., Idelberger S., Herdtfelder M., Heurich M., Kramer-Schadt S., Duchamp C., Drouet-Hoguet N., Morand A., Blanc L., Charbonnel A., Gimenez O. (2022). Eurasian lynx populations in Western Europe: What prospects for the next 50 years? bioRxiv preprint.\nGhislain M., T. Bonnet, U. Godeau, O. Dehorter, O. Gimenez, P.-Y. Henry (2022). Synchrony in adult survival is remarkably strong among common temperate songbirds across France. EcoEvoRxiv preprint.\nLauret V., O. Gimenez, H. Labach, N. Lescureux (2022). Behind the scene of ecological expertise: the place of monitoring for biodiversity workers in the French Mediterranean Marine Protected Areas. SocArXiv preprint.\nViollat L., M. Quéroué, K. Delord, O. Gimenez, C. Barbraud (2023). Bottom-up effects and density dependence drive the dynamic of an antarctic seabird predator-prey system. bioRxiv preprint\nDesprez M., V. Miele, O. Gimenez (2023). Nine tips for ecologists using machine learning. arXiv preprint\nPublished\nMaurer, G., M. Chandelier, B. Mulot, O. Gimenez (2023). Polarized media coverage of conflicting, yet emblematic species: The ambivalent portrayal of the Asian elephant. Biological Conservation 289: 110391. *Co-senior author.\nRigoudy N., A. Benyoub, A. Besnard, C. Birck, Y. Bollet, Y. Bunz, J. Chiffard Carriburu, G. Caussimont, E. Chetouane, P. Cornette, A. Delestrade, N. De Backer, L. Dispan, M. Le Barh, J. Duhayer, J.-F. Elder, J.-B. Fanjul, J. Fonderflick, N. Froustey, M. Garel, W. Gaudry, A. Gérard, O. Gimenez, A. Hemery, A. Hemon, J.-M. Jullien, I. Malafosse, M. Marginean, V. Prunet, J. Rabault, M. Randon, A. Régnier, R. Ribière, J.-C. Ricci, S. Ruette, J. Sentilles, N. Siefert, B. Smith, G. Terpereau, W. Thuiller, A. Uzal, V. Vautrain, G. Dussert, B. Spataro, V. Miele, S. Chamaillé-Jammes (2023). The DeepFaune initiative: a collaborative effort towards the automatic identification of French fauna in camera-trap images. European Journal of Wildlife Research 69: 113.\nBauduin S., O. Gimenez, N. Drouet-Hoguet, J. Louvrier, C. Duchamp (2023). Estimer l’effort d’échantillonnage de réseaux participatifs : l’exemple du réseau Loup-lynx. Biodiversité 5: 6-9.\nNakamura M., J. V. López-Bao, H. Rio-Maior, S. Roque, P. Gil, A. Serronha, E. García, O. Hernández Palacios, G. Ferrão da Costa, F. Álvares, F. Petrucci-Fonseca, O. Gimenez*, P. Monterroso* (2023). Insights into the dynamics of wolf occupancy in human-dominated landscapes. Biological Conservation 286: 110316. *Co-senior author.\nForcada J., Hoffman J. I., Gimenez O., Staniland I. J., Bucktrout P., \u0026amp; Wood A. G. (2023). Ninety years of change, from commercial extinction to recovery, range expansion and decline for Antarctic fur seals at South Georgia. Global Change Biology 00, 1–21.\nLauret V., H. Labach, L. David, M. Authier, O. Gimenez (2023). Using integrated multispecies occupancy models to map co-occurrence between bottlenose dolphins and fisheries in the Gulf of Lion, French Mediterranean Sea. Oikos e10270.\nCouturier T., S. Bauduin, G. Astruc, A. Blanck, C. Canonne, T. Chambert, J. Chiffard, A. Cosquer, S. Cubaynes, L. Curtet, E. Dortel, N. Drouet-Hoguet, C. Duchamp, C. Francesiaz, O. Grente, A. Jailloux, M. Kervellec, V. Lauret, J.-D. Lebreton, J. Louvrier, L. Marescot, R. Mathevet, M.-L. Navas, C. Perrot, N. Poulet, P.-Y. Quenette, M. Salas, G. Souchay, C. Vanpé, A. Besnard, O. Gimenez (2023). Building spaces of interactions between researchers and managers: Case studies with wildlife monitoring and conservation in France. Ecological Solutions and Evidence 4:e12245.\nSilk M., O. Gimenez (2023). Generation and applications of simulated datasets to integrate social network and demographic analyses. Ecology and Evolution 13:e9871.\nKervellec M., C. Milleret, C. Vanpé, P.-Y. Quenette, J. Sentilles, S. Palazón, I. A. Jordana, R. Jato, M. M. Elósegui Irurtia, O. Gimenez (2023). Integrating opportunistic and structured non-invasive surveys with spatial capture-recapture models to map connectivity of the Pyrenean brown bear population. Biological Conservation 278: 109875.\nFernández-López, J., P. Acevedo, O. Gimenez (2023). La unión hace la fuerza: modelos de distribución de especies integrando diferentes fuentes de datos. Ecosistemas 32: 2527-2527.\nRodrigues P., I. Dorresteijn, O. Gimenez (2023). ‘The human shield effect’: Human-wildlife co-occurrence patterns in the coffee forests of southwestern Ethiopia. Food Webs 36: e00288.\nPetridou M., J.F. Benson, O. Gimenez, Y. Iliopoulos, V. Kati (2023). Do husbandry practices reduce depredation of free-ranging livestock? A case study with wolves in Greece. Biological Conservation 283: 11009.\nKleiven E.F., F. Barraquand, Gimenez O., J.-A. Henden, R. Anker Ims, E. M. Soininen, N. G. Yoccoz (2023). A Dynamic Occupancy Model for Interacting Species with Two Spatial Scales. JABES In press.\nGrente O., C. Duchamp, S. Bauduin, S. Chamaille-Jammes, N. Drouet-Hoguet, O. Gimenez (2023). Tirs dérogatoires de loups en France : évaluation des effets sur les dommages aux troupeaux. Paru dans Naturae numéro 5.\nTardy C., D. Ody, O. Gimenez, S. Planes (2023). Abundance of fin whales (Balaenoptera physalus) in the north-western Mediterranean Sea, using photo-identification and microsatellite genotyping. Marine Ecology 44:e12737.\nPetridou M., J.F. Benson, O. Gimenez, V. Kati (2023). Spatiotemporal Patterns of Wolves, and Sympatric Predators and Prey Relative to Human Disturbance in Northwestern Greece. Diversity 15:184.\nFrost F., R. McCrea, R. King, O. Gimenez, E. Zipkin (2023). Integrated Population Models: Achieving Their Potential. Journal of Statistical Theory and Practice 17:6.\nLauret V., H. Labach, D. Turek, S. Laran, O. Gimenez (2022). Integrated spatial models foster complementarity between monitoring programmes in producing large‐scale bottlenose dolphin indicators. Animal Conservation 26: 228-238.\nVanpé C., B. Piédallu, P.-Y. Quenette, J. Sentilles, G. Queney, S. Palazón, I. A. Jordana, R. Jato, M. M. Elósegui Irurtia, J. Solà de la Torre, O. Gimenez (2022). Estimating abundance of a recovering transboundary brown bear population with capture-recapture models. Peer Community Journal 2:e71.\nAdamantopoulou S., A. A. Karamanlidis, P. Dendrinos, O. Gimenez (2022). Citizen science indicates significant range recovery and defines new conservation priorities for Earth\u0026rsquo;s most endangered pinniped in Greece. Animal Conservation 26: 115-125.\nNaciri, M, J Aars, MA Blanchet, O. Gimenez, S Cubaynes (2022). Reproductive senescence in polar bears in a variable environment Frontiers in Ecology and Evolution.\nGimenez O., M. Kervellec, J.-B. Fanjul, A. Chaine, L. Marescot, Y. Bollet, C. Duchamp (2021) Trade-off between deep learning for species identification and inference about predator-prey co-occurrence: Reproducible R workflow integrating models in computer vision and ecological statistics. Computo. https://doi.org/10.57750/yfm2-5f45. Data and code here.\nGrente O., T. Saubusse, Gimenez O., E. Marboutin, C. Duchamp (2022). Wolf depredation hotspots in France: Clustering analyses accounting for livestock availability. Biological Conservation 267: 109495.\nGervasi V., J. Linnell, T. Berce, L. Boitani, B. Cretois, P. Ciucci, C. Duchamp, A. Gastineau, O. Grente, D. Hilfiker, D. Huber, Y. Iliopoulos, A. Karamanlidis, F. Marucco, Y. Mertzanis, P. Mannil, H. Norberg, N. Pagon, L. Pedrotti, P.-Y. Quenette, S. Reljic, V. Salvatori, T. Talvi, M. von Arx, Gimenez O. (2021). Ecological correlates of large carnivore depredation on sheep in Europe. Global Ecology and Conservation 30: e01798.\nKazakou E., Fried G., Cheptou P.-O., and Gimenez O. (2021). Does seed mass drive interspecies variation in the effect of management practices on weed demography? Ecology and Evolution 00: 1– 9.\nMaurer G., Gimenez O., Mulot B., and Lescureux N. (2021). Under pressure: How human-wild-captive elephant social-ecological system in Laos is teetering due to global forces and sociocultural changes. People and Nature 3: 1047-1063.\nRounsevell M.D.A., A. Arneth, C. Brown, W.W.L. Cheung, O. Gimenez, I. Holman, P. Leadley, C. Luján, S. Mahevas, I. Maréchaux, R. Pélissier, P. H. Verburg, G. Vieilledent, B.A. Wintle, Y-J Shin (2021). Identifying uncertainties in scenarios and models of socio-ecological systems in support of decision-making. One Earth 4: 967-985.\nTavecchia G., O. Gimenez, R. Choquet, A. Rotger, D. Oro, S. Tenan, A. Sanz-Aguilar (2021). The trap of hidden processes: Why ‘quick \u0026amp; dirty’methods to estimate mortality are not always good. A comment to De Pascalis et al.(2020). Biological Conservation 256: 109057.\nSantostasi N. L., O. Gimenez, R. Caniglia, E. Fabbri, L. Molinari, W. Reggioni, P. Ciucci (2021). Estimating Admixture at the Population Scale: Taking Imperfect Detectability and Uncertainty in Hybrid Classification Seriously. Journal of Wildlife Management 85: 1031-1046.\nZriki G., R. Blatrix, D. J. Bicout, O. Gimenez, A.‐S. Soulié, L. Dadu, D. Degueldre, G. Chiron, N. Sleeckx, L. Roy (2021). Population‐level impact of native arthropod predators on the poultry red mite Dermanyssus gallinae. Journal of Experimental Zoology Part A: Ecological and Integrative Physiology 335: 552-563.\nMiele V., S. Dray, O. Gimenez (2021) Images, écologie et deep learning.. Regard 95 de la Société Française d\u0026rsquo;Ecologie et d\u0026rsquo;Evolution.\nTurek, D., C. Wehrhahn, Gimenez O. (2021). Bayesian Non-Parametric Detection Heterogeneity in Ecological Models. Environmental and Ecological Statistics 28: 355-381.\nLauret V., Labach H., Authier M., Gimenez O. (2021). Combining single- and repeated-visit occupancy models to make the best of monitoring surveys. Ecology 102: e03535.\nGamelon M, Baubet E., Besnard A., Gaillard J.-M., Lebreton J.-D., Touzot L., Veylit L., Gimenez O. (2021). Efficient use of harvest data: An integrated population model for exploited animal populations. Ecography 44: 1296-1310.\nLabach H., Azzinari C., Barbier M., Cesarini C., Daniel B., David L., Dhermain F., Di-Méglio N., Guichard B., Jourdan J., Robert N., Roul M., Tomasi N., Gimenez O. (2022). Distribution and abundance of bottlenose dolphin (Tursiops truncatus) over the French Mediterranean continental shelf. Marine Mammal Science 38: 212-222.\nBarraquand F., Gimenez O. (2021). Fitting stochastic predator-prey models using both population density and kill rate data Theoretical Population Biology 138: 1-27.\nCubaynes S., Aars J., Yoccoz N.G., Pradel R., Wiig Ø, Ims R.A., Gimenez O. (2021). Modeling the demography of species providing extended parental care: A capture-recapture approach with a case study on Polar Bears (Ursus maritimus). Ecology and Evolution 11: 3380-3392.\nQuéroué M., C. Barbraud, F. Barraquand, D. Turek, K. Delord, N. Pacoureau, Gimenez O. (2021). Multispecies integrated population model reveals bottom-up dynamics in a seabird predator-prey system Ecological Monographs 91: e01459.\nGrente O., C. Duchamp, S. Bauduin, T. Opitz, S. Chamaillé-Jammes, N. Drouet-Hoguet, O. Gimenez (2020) Tirs dérogatoires de loups en France: état des connaissances et des enjeux pour la gestion des attaques aux troupeaux. Paru dans Faune Sauvage numéro 327.\nBischof R., C. Milleret, P. Dupont, J. Chipperfield, M. Tourani, A. Ordiz, P. de Valpine, D. Turek, J. A. Royle, Gimenez O., Ø. Flagstad, M. Åkesson, L. Svensson, H. Brøseth, J. Kindberg (2020). Estimating and forecasting spatial population dynamics of apex predators using transnational genetic monitoring. Proceedings of the National Academy of Sciences 117:30531-30538.\nVitet C., P. Duncan, Gimenez O., C. Mabika, S. Chamaillé‐Jammes (2020). Plains zebras bring evidence that dilution and detection effects may not always matter behaviorally and demographically. Ecosphere 11: e03288.\nBauduin S., O. Grente, N.L. Santostasi, P. Ciucci, C. Duchamp, Gimenez O. (2020). An individual-based model to explore the impacts of lesser-known social dynamics on wolf populations. Ecological Modelling 433: 109209.\nMcClintock B.T., R. Langrock, Gimenez O., E. Cam, D.L. Borchers, R. Glennie, T. A. Patterson (2020). Uncovering ecological state dynamics with hidden Markov models. Ecology Letters 23: 1878-1903.\nGenovart M., Gimenez O., A. Bertolero, R. Choquet, D. Oro, R. Pradel (2020). Decrease in social cohesion in a colonial seabird under a perturbation regime. Scientific reports 10: 18720.\nSantostasi N.L., Ciucci P., Bearzi G., Bonizzoni S., Gimenez O. (2020). Assessing the dynamics of hybridization through a matrix modelling approach. Ecological Modelling 431: 109120.\nChero G., Pradel R., Derville S., Bonneville C., Gimenez O., Garrigue C. (2020). Reproductive capacity of an endangered and recovering population of humpback whales in the Southern Hemisphere. Marine Ecology Progress Series 643: 219-227.\nMilleret C., Dupont P., Chipperfield J., Turek D., Brøseth H., Gimenez O., de Valpine P., Bischof R. (2020). Estimating abundance with interruptions in data collection using open population spatial capture–recapture models. Ecosphere 11( 6):e03172.\nLouvrier L., Papaïx J., Duchamp C., Gimenez O. (2020). A mechanistic-statistical species distribution model to explain and forecast wolf (Canis lupus) colonization in South-Eastern France. Spatial Statistics 36: 100428.\nMarescot L., Lyet A., Singh R., Carter N., Gimenez O. (2019). Inferring wildlife poaching in southeast Asia with multispecies dynamic occupancy models. Ecography 42: 1–12.\nGamble A., Garnier R., Chambert T., Gimenez O., Boulinier T. (2020). Next generation serology: integrating cross-sectional and capture-recapture approaches to infer disease dynamics. Ecology 101:e02923.\nBarraquand F., Gimenez O. (2019). Integrating multiple data sources to fit matrix population models for interacting species. Ecological Modelling 411: 108713.\nGimenez O., Gatti S., Duchamp C., Germain E., Laurent A., Zimmermann F., Marboutin E. (2019). Spatial density estimates of Eurasian lynx (Lynx lynx) in the French Jura and Vosges Mountains. Ecology and Evolution 9: 11707-11715.\nRenner, I.W., Louvrier J., Gimenez O. (2019). Combining multiple data sources in species distribution models while accounting for spatial dependence and overfitting with combined penalized likelihood maximization. Methods in Ecology and Evolution 10: 2118-2128.\nBonnet‐Lebrun A. , Karamanlidis A. A., de Gabriel Hernando M. , Renner I. and O. Gimenez (2019). Identifying priority conservation areas for a recovering brown bear population in Greece using citizen science data. Animal Conservation 23: 83-93\nDonfrancesco V., Ciucci P., [14 authors], Gimenez O., [28 authors] and Mukherjee N. (2019). Unravelling the Scientific Debate on How to Address Wolf-Dog Hybridization in Europe. Frontiers in Ecology and Evolution 7:175. doi: 10.3389/fevo.2019.00175.\nGervasi V., Linnell J.D.C., Brøseth H., Gimenez O. (2019). Failure to coordinate management in transboundary populations hinders the achievement of national management goals: The case of wolverines in Scandinavia. Journal of Applied Ecology 56: 1905-1915.\nLok T., Hassell C.J., Piersma T., Pradel R. Gimenez O. (2019). Accounting for heterogeneity when estimating stopover duration, timing and population size of red knots along the Luannan Coast of Bohai Bay, China. Ecology and Evolution 9: 6176–6188.\nLe Bot T., A. Lescroël, J. Fort, C. Péron, O. Gimenez, P. Provost, D. Grémillet (2019). Fishery discards do not compensate natural prey shortage in Northern gannets from the English Channel. Biological Conservation 236: 375-384.\nDorinda Marie Folio, J. Aars, O. Gimenez, A. E. Derocher, Ø. Wiig and S. Cubaynes (2019). How many cubs can a mum nurse? Maternal age and size influence litter size in polar bears. Biology Letters. 15: 20190070.\nDupont P., Milleret C, Gimenez O., Bischof R (2019). Population closure and the bias‐precision trade‐off in spatial capture–recapture. Methods in Ecology and Evolution 10: 661-672.\nSantostasi N.L., P. Ciucci, R. Caniglia, E. Fabbri, L. Molinari, W. Reggioni, Gimenez O. (2019). Use of hidden Markov capture–recapture models to estimate abundance in the presence of uncertainty: Application to the estimation of prevalence of hybrids in animal populations. * Ecology and evolution * 9: 744-755.\nGimenez O., L. Mansilla, M.J. Klaich, M.A. Coscarella, S.N. Pedraza, E.A. Crespo (2019). Inferring animal social networks with imperfect detection. Ecological Modelling. 401: 69-74.\nMelovski, D., Von Arx, M., Avukatov, V., Breitenmoser-Würsten, C., Đurović, M., Elezi, R., Gimenez O., \u0026hellip; , Breitenmoser, U. (2019). Using questionnaire surveys and occupancy modelling to identify conservation priorities for the Critically Endangered Balkan lynx Lynx lynx balcanicus. Oryx. 54: 706-714.\nBenhaiem S., Marescot L., East M. L., Kramer-Schadt S., Gimenez O., Lebreton J.-D. Hofer H. (2018). Slow recovery from a disease epidemic in the spotted hyena, a keystone social carnivore Communications Biology. 201: 1.\nLouvrier J., Chambert T., Marboutin E., Gimenez O. (2018). Accounting for misidentification and heterogeneity in occupancy studies using hidden Markov models. Ecological Modelling. 387: 61-69.\nSantostasi N. L., Bonizzoni S., Gimenez O., Eddy L., Bearzi G. (2018). Common dolphins in the Gulf of Corinth are Critically Endangered. Aquatic Conserv: Mar Freshw Ecosys. 31: 101-109. https://doi.org/10.1002/aqc.2963.\nLouvrier J., Molinari‐Jobin A., Kéry M., Chambert T., Miller D., Zimmermann F., Marboutin E., Molinari P., Mueller O., Cerne R., Gimenez O. (2019). Use of ambiguous detections to improve estimates from species distribution models. Conservation Biology. 33: 185-195.\nBenhaiem S., Marescot L., Hofer H., East M.L., Lebreton J.-D., Kramer-Schadt S., Gimenez O. (2018). Robustness of eco-epidemiological capture-recapture parameter estimates to variation in infection state uncertainty. Frontiers in Veterinary Science 5: 197.\nGimenez O., Lebreton J.‐D., Choquet R., Pradel R. (2018). R2ucare: An r package to perform goodness‐of‐fit tests for capture–recapture models. Methods in Ecology and Evolution 9: 1749–1754.\nSalguero‐Gomez R., Violle C., Gimenez O., Childs D. (2018). Delivering the promises of trait‐based approaches to the needs of demographic approaches, and vice versa. Functional Ecology 32: 1424–1435.\nBancila R. I., Pradel R., Choquet R., Plaiasu R., Gimenez O. (2018). Using temporary emigration to inform movement behaviour of cave‐dwelling invertebrates: a case study of a cave harvestman species. Ecological Entomology 43: 551-559.\nChandelier M., Steuckardt A., Mathevet R., Diwersy S., Gimenez O. (2018). Content analysis of newspaper coverage of wolf recolonization in France using structural topic modeling Biological Conservationa 220: 254-261.\nMarescot L, Benhaiem S, Gimenez O. et al. (2018). Social status mediates the fitness costs of infection with canine distemper virus in Serengeti spotted hyenas. Functional Ecology 32: 1237–1250.\nBarbraud C., Bertrand A., Bouchón M., Chaigneau A., Delord K., Demarcq H., Gimenez O., Torero M. G., Gutiérrez D., Oliveros-Ramos R., Passuni G., Tremblay Y., Bertrand S. (2018). Density dependence, prey accessibility and prey depletion by fisheries drive Peruvian seabird population dynamics. Ecography 41: 1092-1102.\nMarzolin G., Gimenez O. (2018). Fitting a Gamma-Gompertz survival model to capture-recapture data collected on free-ranging animal populations. Journal of Open Source Software. 3(21): 216.\nBearzi G., Gimenez O. (2018). Searching for meaning in marine mammal shared data. Ethics in Science and Environmental Politics. 18: 9-13.\nHamel S., Gaillard J.-M., Yoccoz N. G., Bassar R. D., Bouwhuis S., Caswell H., Douhard M., Gangloff E. J., Gimenez O., Lee P. C., Smallegange I. M., Steiner U. K., Vedde, O. and Vindenes Y. (2018). General conclusion to the special issue Moving forward on individual heterogeneity. Oikos. 127: 750–756.\nDesprez M., Gimenez O., McMahon C.R., Hindell M.A., Harcourt R.G. (2018). Optimizing lifetime reproductive output: Intermittent breeding as a tactic for females in a long-lived, multiparous mammal. Journal of Animal Ecology. 87: 199–211.\nBumb I., Garnier E., Coq S., Nahmani J., Del Rey Granado M., Gimenez O., E. Kazakou (2018). Traits determining the digestibility– decomposability relationships in species from Mediterranean rangelands. Annals of Botany. 121: 459–469.\nRobardet E., Borel C., Moinet M., Jouan D., Wasniewski M., Barrat J., Boue F., Monchatre-Leroy E., Servat A., Gimenez O., Cliquet F., Picard-Meyer E. (2017). Long-term population surveys of two serotine bat (Eptesicus serotinus) colonies exposed to EBLV-1 (European Bat Lyssavirus type 1): Assessment of rabies transmission using capture-recapture models. PLoS Neglected Tropical Diseases. 11 (11), e0006048.\nGimenez O., E. Cam, J.-M. Gaillard (2018). Individual heterogeneity and capture–recapture models: what, why and how? Oikos. 127: 664–686.\nGimenez O., J.-M. Gaillard (2018). Estimating individual fitness in the wild using capture–recapture data. Population Ecology. 60: 101–109.\nGimenez O., C. Barbraud (2017). Dealing with many correlated covariates in capture–recapture models. Population Ecology. 59: 287-291.\nBertulli C.G., L. Guéry, N. McGinty, A. Suzuki, N. Brannan, T. Marques, M. H. Rasmussen, Gimenez O. (2017). Capture-recapture abundance and survival estimates of three cetacean species in Icelandic coastal waters using trained scientist-volunteers. Journal of Sea Research. 131: 22-31.\nMaurer G., Rashford B. S., Chanthavong V., Mulot B., Gimenez O. (2017). Wild-captive interactions and economics drive dynamics of Asian elephants in Laos. Scientific Reports. DOI:10.1038/s41598-017-13907-x.\nLieury N., Devillard S., Besnard A., Gimenez O., Hameau O., Ponchon C., Millon A. (2017). Designing cost-effective capture-recapture surveys for improving the monitoring of survival in bird populations. Biological Conservation. 214: 233-241.\nTidière M., Lemaître J.-F., Pélabon C., Gimenez O. and Gaillard, J.-M. (2017). Evolutionary allometry reveals a shift in selection pressure on male horn size. Journal of Evolutionary Biology. 30: 1826–1835.\nLagrange P., Gimenez O., Doligez B., Pradel R., Garant D., Pelletier F., Bélisle M. (2017). Assessment of individual and conspecific reproductive success as determinants of breeding dispersal of female tree swallows: A capture-recapture approach. Ecology and Evolution. 7: 7334-7346.\nLouvrier, J., C. Duchamp, V. Lauret, E. Marboutin, S. Cubaynes, R. Choquet, C. Miquel, O. Gimenez (2018). Mapping and explaining wolf recolonization in France using dynamic occupancy models and opportunistic data. Ecography. 41: 647-660.\nPiédallu, B., P.-Y. Quenette, N. Bombillon, A. Gastineau, C. Miquel, O. Gimenez (2019). Determinants and patterns of habitat use by the brown bear Ursus arctos in the French Pyrenees revealed by occupancy modelling. Oryx. 53: 334–343.\nSantostasi, N.L., S. Bonizzoni, G. Bearzi, L. Eddy, O. Gimenez (2016). A Robust Design Capture-Recapture Analysis of Abundance, Survival and Temporary Emigration of Three Odontocete Species in the Gulf of Corinth, Greece. Plos One. 11(12): e0166650.\nKazakou, E., G. Fried, J. Richarte, O. Gimenez, C. Violle and A. Metay (2016). A plant trait-based response-and-effect framework to assess vineyard inter-row soil management. Botany Letters. 163: 373-388.\nPéron, G., J.-M. Gaillard, C. Barbraud, C. Bonenfant, A. Charmantier, R. Choquet, T. Coulson, V. Grosbois, A. Loison, G. Marzolin, N. Owen-Smith, D. Pardo, F. Plard, R. Pradel, C. Toïgo, O. Gimenez (2016). Evidence of reduced individual heterogeneity in adult survival of long-lived species. Evolution. 70: 2909-2914.\nAbadi, F., C. Barbraud, O. Gimene (2017). Integrated population modeling reveals the impact of climate on survival of juvenile emperor penguin. Global Change Biology. 23: 1353-1359.\nMcCrea, R.S., B.J.T. Morgan, O. Gimenez (2017). A new strategy for diagnostic model assessment in capture-recapture. Journal of the Royal Statistical Society: Series C. 66: 815-831.\nTidière, M., J.-M. Gaillard, V. Berger, D.W.H. Müller, L. Bingaman Lackey, O. Gimenez, M. Clauss and J.-F. Lemaître (2016). Comparative analyses of longevity and senescence reveal variable survival benefits of living in zoos across mammals. Scientific Reports. 6: 36361.\nBearzi, G., S. Bonizzoni, N.L. Santostasi, N.B. Furey, L. Eddy, V.D. Valavanis, O. Gimenez (2016). Dolphins in a Scaled-Down Mediterranean: The Gulf of Corinth\u0026rsquo;s Odontocetes. Advances in Marine Biology. 75: 297-331.\nSoldatini, C., Y.V. Albores-Barajas, B. Massa and O. Gimenez (2016). Forecasting ocean warming impacts on seabird demography: a case study on the European storm petrel. Marine Ecology Progress Series. 552: 255–269.\nGervasi, V., Brøseth, H, Gimenez, O., Nilsen, E.B., Odden, J., Flagstad, O. and J.D.C. Linnell (2016). Sharing data improves monitoring of trans-boundary populations: the case of wolverines in central Scandinavia. Wildlife Biology. 22: 95–106.\nPiédallu, B., Quenette, P.-Y., Mounet, C., Lescureux, N., Borelli-Massines, M., Dubarry, E., Camarra, J.-J. \u0026amp; Gimenez, O. (2016). Spatial variation in public attitudes towards brown bears in the French Pyrénées. Biological Conservation. 197: 90–97.\nBoulenger, C., A. Acou, O. Gimenez, F. Charrier, J. Tremblay \u0026amp; E. Feunteun (2016). Factors underlying survival of yellow European eels in two unexploited eel sub-populations. Freshwater Biology. 61: 947-962.\nGager, Y., O. Gimenez, M. Teage O\u0026rsquo;Mara and Dina K.N. Dechmann (2016) Group size, survival and surprisingly short lifespan in socially foraging bats. BMC Ecology. 16:2.\nC.A. Bost, C. Cotté, P. Terray, C. Barbraud, C. Bon, K. Delord, O. Gimenez, Y. Handrich, Y. Naito, C. Guinet, H. Weimerskirch (2015). Large-scale climatic anomalies affect marine predator foraging behaviour and demography. Nature Communications. 6: 8220.\nTidière, M., J.-M. Gaillard, D.W.H. Müller, L. Bingaman Lackey, O. Gimenez, M. Clauss, J.-F. Lemaître (2015). Does sexual selection shape sex-differences in longevity and senescence patterns across vertebrates? A review and new insights from captive ruminants. Evolution. 69: 3123–3140 .\nMouquet, N., Lagadeuc, Y., Devictor, V., Doyen, L., Duputié, A., Eveillard, D., Faure, D., Garnier, E., Gimenez, O., Huneman, P., Jabot, F., Jarne, P., Joly, D., Julliard, R., Kéfi, S., Kergoat, G. J., Lavorel, S., Le Gall, L., Meslin, L., Morand, S., Morin, X., Morlon, H., Pinay, G., Pradel, R., Schurr, F. M., Thuiller, W., Loreau, M. (2015). Predictive ecology in a changing world. Journal of Applied Ecology. 52: 1293–1310.\nGamelon, M., Gaillard, J.-M., Gimenez, O., Coulson, T., Tuljapurkar, S., Baubet, E. (2016). Linking demographic responses and life history tactics from longitudinal data in mammals. *Oikos. *125: 395–404.\nKaramanlidis, A.A., M. de Gabriel Hernando, L. Krambokoukis, O. Gimenez (2015). Evidence of a large carnivore population recovery: counting bears in Greece. Journal for Nature Conservation. 27: 10-17.\nBischof, R., H. Broseth, O. Gimenez (2016). Wildlife in a politically divided world: insularism inflates estimates of brown bear abundance. Conservation Letters. 9: 122–130.\nGimenez, O., Buckland, S.T., Morgan, B.J.T., Bez, N., Bertrand, S., Choquet, R., Dray, S., Etienne, M.-P., Fewster, R., Gosselin, F., Mérigot, B., Monestiez, P., Morales, J., Mortier, F., Munoz, F., Ovaskainen, O., Pavoine, S., Pradel, R., Schurr, F.M., Thomas, L., Thuiller, W., Trenkel, V., de Valpine P., E. Rexstad (2015). Statistical ecology comes of age. Biology Letters. 10: 20140698. Data and R codes.\nTidière, M., J.-M. Gaillard, D.W.H. Müller, L. Bingaman Lackey, O. Gimenez, M. Clauss, J.-F. Lemaître (2014). Males do not senesce faster in large herbivores with highly seasonal rut. Experimental Gerontology. 60: 167–172.\nKoons, D.N., F. Colchero, K. Hersey, and O. Gimenez (2015). Disentangling the effects of climate, density dependence, and harvest on an iconic large herbivore\u0026rsquo;s population dynamics. Ecological Applications 25: 956–967.\nGervasi, V., Brøseth, H., Gimenez, O., Nilsen, E., Linnell, J. (2014). The risks of learning: confounding detection and demographic trend when using count-based indices for population monitoring. Ecology and Evolution. 4: 4637–4648.\nGamelon, M., Gimenez, O., Baubet, E., Coulson, T., Tuljapurkar, S., Gaillard, J.-M. (2014). Influence of Life-History Tactics on Transient Dynamics: A Comparative Analysis across Mammalian Populations. The American Naturalist 184: 673-683.\nBarnagaud, J.Y., Papaïx, J., Gimenez, O., Svenning, J.-C. (2014). Dynamic spatial interactions between the native invader Brown-headed Cowbird and its hosts - a metacommunity approach. Diversity and Distributions. 21: 511-522.\nOrgeret, F, Garrigue, C., Gimenez, O., Pradel, R. (2014). Robust assessment of population trends in marine mammals applied to the New Caledonia Humpback Whales. Marine Ecology Progress Series 515: 265–273.\nKoons, D.N., M. Gamelon, J.-M. Gaillard, L.M. Aubry, R.F. Rockwell, F. Klein, R. Choquet and O. Gimenez (2014). Methods for Studying Cause-Specific Senescence in the Wild. Methods in Ecology and Evolution 5: 924-933.\nGamelon, M., S. Focardi, J.-M. Gaillard, O. Gimenez, C. Bonenfant, B. Franzetti, R. Choquet, F. Ronchi, E. Baubet, J.-F. Lemaître. (2014). Do age-specific survival patterns of wild boar fit current evolutionary theories of senescence? Evolution. 68: 3636–3643.\nBlanc, L., E. Marboutin, S. Gatti, F. Zimmermann and O. Gimenez (2014). Improving abundance estimation by combining capture-recapture and occupancy data: example with a large carnivore. Journal of Applied Ecology. 51: 1733–1739.\nGimenez, O., L. Blanc, A. Besnard, R. Pradel, P. F. Doherty Jr, E. Marboutin and R. Choquet (2014). Fitting occupancy models with E-SURGE: hidden-Markov modelling of presence-absence data. Methods in Ecology and Evolution. 5: 592–597. Wiki.\nLee, K.A., C. Huveneers, O. Gimenez, V. Peddemors, R.G. Harcourt (2014). To catch or to sight? A comparison of demographic parameter estimates obtained from mark-recapture and mark-resight models. *Biodiversity and Conservation.*23: 2781-2800.\nSoldatini, C., Y. V. Albores-Barajas, B. Massa and O. Gimenez (2014). Climate Driven Life Histories: The Case of the Mediterranean Storm Petrel. Plos One 9: e94526.\nSantin-Janin, H., B. Hugueny, P. Aubry, D. Fouchet, O. Gimenez and D. Pontier (2014). Accounting for Sampling Error When Inferring Population Synchrony from Time-Series Data: A Bayesian State-Space Modelling Approach with Applications. Plos One 9: e87084.\nHardouin, L. A., A. Robert, M. Nevoux, O. Gimenez, F. Lacroix and Y. Hingrat (2014). Meteorological conditions influence short-term survival and dispersal in a reinforced bird population. Journal of Applied Ecology. 51: 1494–1503.\nDesprez M., Harcourt R., Hindell M.A., Cubaynes S., Gimenez O., McMahon C.R. (2014). Age-specific cost of first reproduction in female southern elephant seals. Biology Letters 10: 20140264.\nCubaynes, S. C. Lavergne, O. Gimenez (2014). Fitting animal survival models with temporal random effects. Environmental and Ecological Statistics* *21: 599–610. Codes.\nLagrange, P., R. Pradel, M. Bélisle, O. Gimenez (2014). Estimating dispersal among numerous sites using capture-recapture data. Ecology. 95: 2316-2323.\nSouchay, G., O. Gimenez, G. Gauthier, R. Pradel (2014). Variations in Band Reporting Rate and Implications for Kill Rate in Greater Snow Geese. Avian Conservation and Ecology. 9: 1.\nLe Saout, S., S. Chollet, S. Chamaille-Jammes, L. Blanc, S. Padié, T. Verchère, J.A. Gaston, M. Gillingham, O. Gimenez, K.L. Parker, D. Picot, H. Verheyden, J.-L. Martin (2014). Understanding the Paradox of Deer Persisting at High Abundance in Heavily Browsed Habitats. Wildlife Biology. 20: 122-135.\nMillon, A., S.J. Petty, B. Little, O. Gimenez, T. Cornulier, X. Lambin (2014). Deleterious effect of dampening prey cycle overrides the impact of climate change on predator dynamics: a long term demographic study on tawny owls. Global Change Biology. 20: 1770–1781.\nCole, D., B.J.T. Morgan, R.S. McCrea, R. Pradel, O. Gimenez, R. Choquet (2014). Does Your Species Have Memory? Analysing Capture-Recapture Data with Memory Models. Ecology and Evolution. 4: 2124–2133.\nAbadi, F., C. Barbraud, D. Besson, J. Bried, P.-A. Crochet, K. Delord, J. Forcada, V. Grosbois, R. A. Phillipse, P. Sagar, P. Thompson, S. Waugh, H. Weimerskirch, A.G. Wood, O. Gimenez (2014). Importance of accounting for phylogenetic dependence in multi-species mark–recapture studies. Ecological Modelling. 273: 236-241.\nFrederiksen, M., Lebreton, J.-D., Pradel, R., Choquet, R. and O. Gimenez (2014). Identifying links between vital rates and environment: a toolbox for the applied ecologist. Journal of Applied Ecology. 51: 71-81.\nMarescot, L., Chapron, G., Chadès, I., Fackler, P., Duchamp, C., Marboutin, E. and O. Gimenez (2013). Complex decisions made simple: A primer on stochastic dynamic programming. Methods in Ecology and Evolution 4: 872-884. Codes.\nGamelon, M., M. Douhard, E. Baubet, O. Gimenez, S Brandt and J.-M. Gaillard (2013). Fluctuating food resources influence developmental plasticity in wild boar. Biology Letters 9: 20130419.\nDesprez, M., McMahon, C., Hindell, M., Harcourt, R. and O. Gimenez (2013). Known unknowns in an imperfect world: incorporating uncertainty in recruitment estimates using multi-event capture-recapture models. Ecology and Evolution 3: 4658-4668.\nDesprez, M., Crivelli, A.J., Lebel I., Massez G. and O. Gimenez (2013). Demographic assessment of a stocking experiment in European Eels. Ecology of Freshwater Fish 22: 412-420.\nLamy, T., O. Gimenez, J.-P. Pointier, P. Jarne and P. David (2013). Metapopulation dynamics of species with cryptic life stages. The American Naturalist 181: 479-491. Codes.\nMadon, B., C. Garrigue, R. Pradel, O. Gimenez (2013). Transience in the humpback whale population of New Caledonia and implications for abundance estimation. Marine Mammal Science 29: 669-678. Codes.\nGamelon, M., J.-M. Gaillard, E. Baubet, S. Devillard, L. Say, S. Brandt \u0026amp; O. Gimenez (2013). The relationship between phenotypic variation among offspring and mother body mass in wild boar: Evidence of coin-flipping? Journal of Animal Ecology 82: 937-945.\nChoquet, R., A. Sanz-Aguilar, B. Doligez, E. Nogué, R. Pradel, L. Gustafsson and O. Gimenez (2013). Estimating demographic parameters from capture-recapture data with dependence among individuals within clusters. Methods in Ecology and Evolution 4: 474-482.\nPéron, G., Y. Ferrand, G. Leray and O. Gimenez (2013). Waterbird demography as indicator of wetland health: The French-wintering common snipe population. Biological Conservation. 164: 123-128\nBolker, B., Gardner, B., Maunder, M., Berg, C., Brooks, M., Comita, L., Crone, E., Cubaynes, S., Davies, T., de Valpine, P., Ford, J., Gimenez, O., Kéry, M., Kim, E., Lennert-Cody, C., Magnusson, A., Martell, S., Nash, J., Nielsen, A., Regetz, J., Skaug, H., Zipkin, E. (2013). Strategies for fitting nonlinear ecological models in R, AD Model Builder, and BUGS. Methods in Ecology and Evolution 4: 501-512. Website.\nPraud, A., O. Gimenez, G. Zanella, N. Pozzi, V. Antras, L. Meyer, B. Garin-Bastuji (2013). Evaluation of five serological tests for the diagnosis of porcine brucellosis in French Polynesia. Trop Anim Health Prod 45: 933-936\nBellan, S., O. Gimenez, R. Choquet and W. Getz (2012). A Hierarchical Distance Sampling Approach to Estimating Mortality Rates from Opportunistic Carcass Surveillance Data. Methods in Ecology and Evolution 4: 361-369\nde Villemereuil, P., Gimenez, O., Doligez, B. (2012). Comparing parent-offspring regression with Frequentist and Bayesian animal models to estimate heritability in wild populations: a simulation study for Gaussian and binary traits. Methods in Ecology and Evolution 4: 260-275. Codes.\nCam, E., O. Gimenez, R. Alpizar-Jara, L. M. Aubry, M. Authier, E. G. Cooch, D. N. Koons, W. A. Link, J.-Y. Monnat, J. D. Nichols, J. J. Rotella, J. A. Royle, R. Pradel (2013). Looking for a needle in a haystack: inference about individual fitness components in a heterogeneous population. Oikos 122: 739-753.\nLebreton, J.-D. and O. Gimenez (2013). Detecting and estimating density-dependence in wildlife populations. Journal of Widlife Management 77: 12-23.\nGimenez, O., Abadi, F., Barnagaud, J.-Y., Blanc, L., Buoro, M., Cubaynes, S., Desprez, M., Gamelon, M., Guilhaumon, F., Lagrange, P., Madon, B., Marescot, L., Papadatou, E., Papaïx, J., Péron, G., Servanty, S. (2013). How can quantitative ecology be attractive to young scientists? Balancing computer/desk work with field work. Animal Conservation 16: 134-136.\nBergeron, P., Montiglio, P.-O., Reale, D., Humphries, M., Gimenez, O. and D. Garant (2013). Disruptive viability selection on adult exploratory behavior in eastern chipmunks. Journal of Evolutionary Biology 26: 766-774.\nBlanc, L., E. Marboutin, S. Gatti, O. Gimenez (2013). Abundance of rare and elusive species: empirical investigation of closed vs. spatially explicit capture-recapture models with Lynx as a case study. Journal of Wildlife Management 77: 372-378.\nPettex, E., S.-H. Lorentsen, D. Grémillet, O. Gimenez, R.T. Barrett, J.-B. Pons, C. Le Bohec and F. Bonadonna (2012). Multi-scale foraging variability in northern gannet (Morus bassanus) fuels potential foraging plasticity. Marine Biology 159: 2743-2756\nBauduin, S., J. Martin, H.H. Edwards, O. Gimenez, S. Koslovsky and D. Fagan (2012). An index of risk of co-occurrence between marine mammals and watercrafts: example of the Florida manatee. Biological Conservation 159: 127-136\nCharpentier, M.J.E., Huchard, E., Widdig, A., Gimenez, O., Sallé, B., Kappelee, P. and J.P. Renoult (2012). Distribution of affiliative behavior accross kin classes and their fitness consequences in mandrills. Ethology 118: 1198-1207\nGimenez, O., Lebreton, J.-D., Gaillard, J.-M., Choquet, R. and R. Pradel (2012). Estimating demographic parameters using hidden process dynamic models. Theoretical Population Biology 82: 307-316. Codes.\nBuoro, M., E. Prévost, O. Gimenez (2012). Digging through model complexity: using hierarchical models to uncover evolutionary processes in the wild. Journal of Evolutionary Biology 25: 2077-2090\nHardouin, L. A., M. Nevoux, A. Robert, O. Gimenez and Y. Hingrat (2012). Determinisms and costs of natal dispersal in a lekking species. Oïkos 121: 804-812\nGamelon, M., Gaillard, J.-M., S. Servanty, O. Gimenez, C. Toïgo, E. Baubet, F. Klein, J.-D. Lebreton (2012). Making use of harvest information to examine alternative management scenarios: a body weight-structured model for wild boar. Journal of Applied Ecology 49: 833-841.\nLibois, E., O. Gimenez, D. Oro, E. Mínguez, R. Pradel, A. Sanz-Aguilar (2012). Nest boxes: A successful management tool for the conservation of an endangered seabird. Biological Conservation 155: 39-43.\nAbadi, F., O. Gimenez, H. Jakobere, W. Stauberf, R. Arlettaz and M. Schaub (2012). Estimating the strength of density dependence in the presence of observation errors using integrated population models. Ecological Modelling 242: 1-9.\nLebreton, J.-D., R. Choquet and O. Gimenez (2012). Simple estimation and test procedures in capture-mark-recapture mixed models. Biometrics 68: 494-503.\nCubaynes, S., C. Lavergne, E. Marboutin, and O. Gimenez (2012). Assessing individual heterogeneity using model selection criteria: How many mixture components in capture-recapture models? Methods in Ecology and Evolution 3: 564-573. Codes.\nFrank, B., O. Gimenez and P. Baret (2012). Assessing brown trout spawning movements with multistate capture-recapture models: a case study in a fully-controlled Belgian brook. Canadian Journal of Fisheries and Aquatic Sciences 69: 1091-1104\nPéron, G., Y. Ferrand, F. Gossmann, C. Bastat, M. Guenezan and O. Gimenez (2012). Spatial heterogeneity in mortality and its impact on the population dynamics of Eurasian woodcocks. Population Ecology 54: 305-312\nCubaynes, S, C. Doutrelant, A. Gregoire, P. Perret, B. Faivre, and O. Gimenez (2012). Testing hypotheses in evolutionary ecology with imperfect detection: Structural equation modeling of mark-recapture data. Ecology 93: 248-255. Codes.\nGimenez, O., T. Anker-Nilssen and V. Grosbois (2012). Exploring causal pathways in demographic parameter variation: Path analysis of mark-recapture data. Methods in Ecology and Evolution 3: 427-432. Codes.\nChoquet, R. and O. Gimenez (2012). Towards built-in capture-recapture mixed models in program E-SURGE. Journal of Ornithology 152: 625-639.\nMarescot, L., O. Gimenez, C. Duchamp, E. Marboutin and G. Chapron (2012). Reducing matrix population models with application to social animal species. Ecological Modelling 232: 91-96\nPapadatou, E., R. Pradel, M. Schaub, D. Dolch, H. Geiger, C. Ibañez, G. Kerth, A. Popa-Lisseanu, W. Schorcht, J. Teubner and O. Gimenez (2012). Comparing survival among species with imperfect detection using multilevel analysis of mark-recapture data: a case study on bats. Ecography 35: 153-161. Codes.\nPraud, A., O. Gimenez, G. Zanella, B. Dufour, N. Pozzi, V. Antras, L. Meyer, B. Garin-Bastuji (2012). Estimation of sensitivity and specificity of five serological tests for the diagnosis of porcine brucellosis. Preventive Veterinary Medicine 104: 94-100\nArchaux, F., P.-Y. Henry and O. Gimenez (2012). When can we ignore the problem of imperfect detection in comparative studies? Methods in Ecology and Evolution 3: 188-194. Codes.\nCaniglia, R., E. Fabbri, S. Cubaynes, O. Gimenez, J.-D. Lebreton and E. Randi (2012). An improved procedure to estimate wolf abundance using non-invasive genetic sampling and capture-recapture mixture models. Conservation Genetics 13: 53-64\nFletcher, D., Lebreton, J.-D., Marescot, L., Schaub, M., Gimenez, O., Slooten, E. and S. Dawson (2012). Bias in estimation of adult survival and asymptotic population growth rate caused by undetected capture heterogeneity. Methods in Ecology and Evolution 3: 206-216\nBuoro, M., O. Gimenez and E. Prévost (2011). Assessing adaptive phenotypic plasticity by means of conditional strategies from empirical data: The latent environmental threshold model. Evolution 66: 996-1009. Codes.\nGamelon, M., A. Besnard, J.-M. Gaillard, S. Servanty, E. Baubet, S. Brandt and O. Gimenez (2011). High hunting pressure selects for earlier birth data: Wild boar as a case study. Evolution 65: 3100-3112\nMarescot, L., Pradel, R., Duchamp, C. Cubaynes, S., Marboutin, E., Choquet, R. Miquel, C. and O. Gimenez (2011). Capture-recapture population growth rate as a robust tool against detection heterogeneity for population management. Ecological Applications 21: 2898-2907\nPéron, G., Y. Ferrand, F. Gossmann, C. Bastat, M. Guenezan and O. Gimenez (2011). Escape migration decisions in Eurasian Woodcocks: insights from survival analyses using large-scale recovery data. Behavioral Ecology and Sociobiology 65: 1949-1955\nCharmantier, A., M. Buoro, O. Gimenez and H. Weimerskirch (2011). Heritability of short-scale natal dispersal in a large-scale foraging bird, the wandering albatross. Journal of Evolutionary Biology 24: 1487-1496. Codes.\nDesprez, M. R. Pradel, E. Cam, J.-Y. Monnat and O. Gimenez (2011). Now you see him, now you don\u0026rsquo;t: Experience, not age, is related to reproduction in Kittiwakes. Proceedings of the Royal Society B: Biological Sciences 278: 3060-3066.\nMadon, B., O. Gimenez, B. McArdle, C. Baker and C. Garrigue (2011). A New Method For Estimating Animal Abundance with Two Sources of Data in Capture-Recapture Studies. Methods in Ecology and Evolution 2: 390-400\nNevoux, M., O. Gimenez, M. Nicoll, D. Arlt, C. Jones and K. Norris (2011). Population regulation of territorial species: both site-dependence and interference mechanisms matter. Proceedings of the Royal Society B: Biological Sciences 278: 2173-2181\nPéron, G., Y. Ferrand, F. Gossmann, C. Bastat, M. Guenezan and O. Gimenez (2011). Nonparametric spatial regression of survival probability: visualization of population sinks in Eurasian Woodcock. Ecology 92: 1672-1679\nServanty, S., J.-M. Gaillard, Ronchi, F., Focardi, S., Baubet E. \u0026amp; O. Gimenez (2011). Influence of harvesting pressure on demographic tactics: implications for wildlife management. Journal of Applied Ecology 48: 835–843\nFontaine, R., O. Gimenez and J. Bried (2011). The impact of introduced predators, light-induced mortality of fledglings and poaching on the dynamics of the Cory’s shearwater (Calonectris diomedea) population from the Azores, northeastern subtropical Atlantic. Biological Conservation 144: 1998–2011\nMarzolin, G., A. Charmantier and O. Gimenez (2011). Frailty in state-space models: application to actuarial senescence in the dipper. Ecology 92: 562-567. Codes.\nCubaynes, S., P. F. Doherty Jr, E. A. Schreiber, R. W. Schreiber and O. Gimenez (2011). To breed or not to breed: seabirds response to extreme climatic events. Biology Letters 7: 303-306.\nPapadatou, E. Ibáñez, C., Pradel, R. Juste, J. and O. Gimenez (2011). Assessing survival in a multi-population system: a case study on bat populations. Oecologia 165: 925-933. Codes.\nBarnagaud, J.-Y., Crochet, P.-A., Magnani, Y., Bernard-Laurent, A., Menoni, E., Novoa, C. and O. Gimenez (2011). Short-term response to the North Atlantic Oscillation but no long-term effects of climate change on the reproductive success of an alpine bird. Journal of Ornithology 152: 631-641.\nMcCrea, R. S., B.J.T. Morgan, O. Gimenez, P. Besbeas, J.-D. Lebreton and T. Bregnballe (2010). Multi-site Integrated Population Modelling. Journal of Agricultural, Biological, and Environmental Statistics 15: 539-561.\nPapaïx, J., S. Cubaynes, M. Buoro, A. Charmantier, P. Perret and O. Gimenez (2010). Combining capture–recapture data and pedigree information to assess heritability of demographic parameters in the wild. Journal of Evolutionary Biology 23: 2176-2184. Codes.\nServanty, S., R. Choquet, E. Baubet, S. Brandt, J.-M. Gaillard, M. Schaub, C. Toïgo, J.-D. Lebreton, M. Buoro and O. Gimenez (2010). Assessing whether mortality is additive using marked animals: a Bayesian state-space modeling approach. Ecology 91: 1916–1923. Codes.\nBesnard, A., Novoa, C. and O. Gimenez (2010). Hunting impact on Pyrenean gray partridge population dynamics. Wildlife Biology 16: 135-143.\nBuoro, M., Prévost, E. and O. Gimenez (2010). Investigating evolutionary trade-offs in wild populations of Atlantic salmon (Salmo salar): incorporating detection probabilities and individual heterogeneity. Evolution 64: 2629-2642.\nSchaub, M. Aebischer, A. Gimenez, O., Berger, S. and R. Arlettaz (2010). Massive immigration balances high anthropogenic induced mortality in a stable eagle owl population: lessons for conservation. Biological Conservation 143: 1911–1918.\nGuilhaumon, F., Mouillot, D. and O. Gimenez (2010). mmSAR: an R-package for multimodel species-area relationship inference. Ecography 33: 420-424.\nPéron, G. Gimenez, O., Charmantier, A., Gaillard, J.-M. and P.-A. Crochet (2010). Age at the onset of senescence in birds and mammals is predicted by early-life performance. Proceedings of the Royal Society B: Biological Sciences 277: 2849–2856.\nGimenez, O. and R. Choquet (2010). Incorporating individual heterogeneity in studies on marked animals using numerical integration: capture-recapture mixed models. Ecology 91: 951-957.\nPéron, G., Crochet, P.A.C., Choquet, R., Pradel, R., Lebreton, J.-D. and O. Gimenez. (2010). Capture-recapture models with heterogeneity to study survival senescence in the wild. Oïkos 119: 524-532. Appendix.\nCubaynes, S. Pradel, R. Choquet, R. Duchamp, C. Gaillard, J.-M., Lebreton, J.-D., Marboutin, E., Miquel, C., Reboulet, A.-M., Poillot, C., Taberlet, P. and O. Gimenez. (2010). Importance of accounting for detection heterogeneity when estimating abundance: the case of French wolves. Conservation Biology 24: 621-626.\nAbadi, F., Gimenez, O., Arlettaz, R. and M. Schaub (2010). An assessment of integrated population models: bias, accuracy, and violation of the assumption of independence. Ecology 91: 7-14.\nAbadi, F., Gimenez, O., Ullrich, B., Arlettaz, R. and M. Schaub (2010). Estimation of immigration rate using integrated population models. Journal of Applied Ecology 47: 393-400.\nMihoub, J.-B., Gimenez, O., Pilard, P. and F. Sarrazin (2010). Challenging conservation of migratory species: Sahelian rainfalls drive first-year survival of the vulnerable Lesser Kestrel Falco naumanni. Biological Conservation 143: 839-847.\nGimenez, O., Grégoire, A. and T. Lenormand (2009). Estimating and visualizing fitness surfaces using mark-recapture data. Evolution. 63: 3097-3105. Codes.\nChoquet, R., Lebreton, J.-D., Gimenez, O., Reboulet, A.-M., and R. Pradel. (2009). U-CARE: Utilities for performing goodness of fit tests and manipulating CApture-REcapture data. Ecography. 32: 1071-1074.\nKazakou, E., Violle, C, Roumet, C., Pintor, C., Gimenez, O. and E. Garnier. (2009). Litter quality and decomposability of species from a Mediterranean succession depend on leaf traits but not on nitrogen supply. Annals of Botany 104: 1151-1161\nGrosbois, V., Harris, M. P., Anker-Nilssen, T., McCleery, R. H., Shaw, D. N., Morgan, B.J.T. and O. Gimenez (2009). Modeling survival at multi-population scales using mark-recapture data. Ecology 90: 2922-2932. Codes.\nGuilhaumon, F., O. Gimenez, K. J. Gaston, and D. Mouillot (2008). Reply to Bode and Murdoch: A proper integration of species-area relationship uncertainties into return on investment analyses is needed. Proceedings of the National Academy of Sciences. 106: E13\nBischof, R., J. E. Swenson, N. G. Yoccoz, A. Mysterud, and O. Gimenez (2009). The magnitude and selectivity of natural and multiple anthopogenic mortality causes in hunted brown bears. Journal of Animal Ecology 78: 656-665.\nMennerat, A., P. Perret, P. Bourgault, J. Blondel, O. Gimenez, D. W. Thomas, P. Heeb, M. M. Lambrechts (2009). Aromatic plants in nests of blue tits: positive effects on nestlings. Animal Behaviour 77: 569–574.\nGimenez, O., S. Bonner, R. King, R. A. Parker, S.P. Brooks, L. E. Jamieson, V. Grosbois, B. J. T. Morgan, and L. Thomas (2009). WinBUGS for Population Ecologists: Bayesian Modeling Using Markov Chain Monte Carlo Methods. In Modeling Demographic Processes in Marked Populations. Thomson, D. L.; Cooch, E. G.; Conroy, M. J. (Eds.). Springer Series: Environmental and Ecological Statistics, Vol. 3, pages 883-915. Codes.\nGimenez, O., Morgan, B.J.T. and S.P. Brooks (2009). Weak Identifiability in Models for Mark-Recapture-Recovery Data. In Modeling Demographic Processes in Marked Populations. Thomson, D. L.; Cooch, E. G.; Conroy, M. J. (Eds.). Springer Series: Environmental and Ecological Statistics, Vol. 3, pages 1055-1067. Codes.\nGimenez, O. and C. Barbraud (2009). The Efficient Semiparametric Regression Modeling of Capture-Recapture Data: Assessing the Impact of Climate on Survival of Two Antarctic Seabird Species. In Modeling Demographic Processes in Marked Populations. Thomson, D. L.; Cooch, E. G.; Conroy, M. J. (Eds.). Springer Series: Environmental and Ecological Statistics, Vol. 3, pages 43-58.\nBorysiewicz, R.S., B.J.T. Morgan, V. Hénaux, T. Bregnballe, J-D. Lebreton, and O. Gimenez (2009). An Integrated Analysis of Multisite Recruitment, Mark-Recapture- Recovery and Multisite Census Data. In Modeling Demographic Processes in Marked Populations. Thomson, D. L.; Cooch, E. G.; Conroy, M. J. (Eds.). Springer Series: Environmental and Ecological Statistics, Vol. 3, pages 579-591.\nGuilhaumon, F. O. Gimenez, K. J. Gaston, and D. Mouillot (2008). Taxonomic \u0026amp; regional uncertainty in species-area relationships and the identification of richness hotspots. Proceedings of the National Academy of Sciences. 105: 15458–15463.\nGimenez, O., A. Viallefont, A. Charmantier, R. Pradel, E. Cam, C. R. Brown, M. D. Anderson, M. Bomberger Brown, R. Covas, J.-M. Gaillard (2008). The risk of flawed inference in evolutionary studies when detectability is less than one. The American Naturalist. 172: 441–448.\nGimenez,O. (2008). Discussion: Towards a Bayesian analysis template? The Canadian Journal of Statistics. 36: 21-24.\nGrosbois V., Gimenez O., Gaillard, J.-M., Pradel, R., Barbraud, C., Clobert, J., Møller, A. P. and H. Weimerskirch (2008). Assessing the impact of climate variation on survival in vertebrate populations. Biological Reviews 83: 357-399.\nPradel, R., L. Maurin-Bernier, O. Gimenez, M. Genovart, R. Choquet, and D. Oro (2008). Estimation of sex-specific survival with uncertainty in sex assessment. The Canadian Journal of Statistics 36: 29-42.\nCharpentier, M.J.E., F. Prugnolle, O. Gimenez and A. Widdig (2008). Genetic heterozygosity and sociality in a primate species. Behavior Genetics 38: 151-158.\nGhimire, S., Gimenez, O., Pradel, R., McKey, D., and Y. Aumeeruddy-Thomas (2008). Demographic variation and population viability in a threatened Himalayan medicinal and aromatic herb (Nardostachys grandiflora): matrix modelling of harvesting effects in two contrasting habitats. Journal of Applied Ecology 45: 41-51.\nVéran, S., Gimenez, O., Flint, E., Kendall, B., Doherty, P., and J.-D. Lebreton (2007). Quantifying the impact of longline fisheries on adult survival in the Black-footed Albatross. Journal of Applied Ecology 44: 942–952.\nGimenez, O., V. Rossi, R. Choquet, C. Dehais, B. Doris, H. Varella, J.-P. Vila and R. Pradel (2007). State-space modelling of data on marked individuals. Ecological Modelling 206: 431-438. Codes.\nSchaub, M., Gimenez, O., Sierro, A. and R. Arlettaz (2007). Use of Integrated Modeling to Enhance Estimates of Population Dynamics Obtained from Limited Data. Conservation Biology 21: 945–955. Codes.\nKazakou, E., E. Garnier and Gimenez, O. (2007). Contribution of leaf lifespan and nutrient resorption to mean residence time: an elasticity analysis. Ecology 88: 1857-1863.\nPinay, G., B. Gumiero, E. Tabacchi, O. Gimenez, A.M. Tabacchi-Planty, M.M. Hefting, T.P. Burt, V.A. Black, C. Nilsson, V. Iordache, F. Bureau, L. Vought, G.E. Petts and H. Décamps. (2007). Patterns of denitrification rates in European alluvial soils under various hydrological regimes. Freshwater Biology 52: 252–266.\nGimenez O., C. Crainiceanu, C. Barbraud, S. Jenouvrier and B.J.T. Morgan. (2006). Semiparametric regression in capture-recapture modelling. Biometrics 62: 691-698. Codes.\nGimenez, O., R. Covas, C. R. Brown, M. D. Anderson, M. Bomberger Brown and T. Lenormand. (2006). Nonparametric estimation of natural selection on a quantitative trait using mark-recapture data. Evolution 60: 460-466. Codes.\nCharpentier M., Peignot P., Hossaert-McKey M., Gimenez O., Setchell J. and E.J. Wickings (2005). Constraints on control: factors influencing reproductive success in male mandrils (Mandrillus sphinx). Behavioral Ecology 16: 614-623.\nDo F.C., Goudiaby V.A., Gimenez O., Diagne A.L., Diouf M., Rocheteau A., Akpo L.E. (2005) Environmental influence on canopy phenology in the dry tropics Forest Ecology and Management. 215: 319-328.\nGimenez O., Choquet R., Lamor L., Scofield P., Fletcher D., J.-D. Lebreton and R. Pradel (2005). Efficient Profile-Likelihood Confidence Intervals for Capture-Recapture Models. Journal of Agricultural, Biological, and Environmental Statistics 10: 184-196.\nPradel, R., Gimenez O. and J.-D. Lebreton (2005). Principles and interest of GOF tests for multistate capture-recapture models. Animal Biodiversity and Conservation 28: 189–204.\nAncrenaz M., Gimenez O., Ambu L., Ancrenaz K., Andau P., Goossens B., Payne J., Tuuga A. and I. Lackman- Ancrenaz (2005). Aerial Survey Gives New Estimates for Orang-utans in Sabah, Malaysia. Plos Biology 3: 30-37.\nGimenez O., Viallefont A., Choquet R., Catchpole E.A. and B.J.T. Morgan (2004). Methods for investigating parameter redundancy. Animal Biodiversity and Conservation 27: 561-572.\nChoquet R., Reboulet A.-M., Pradel R., Gimenez O. and J.-D. Lebreton (2004). M-SURGE: new software specically designed for multistate capture-recapture models. Animal Biodiversity and Conservation 27: 207-215.\nSchaub M., Gimenez O., Schmidt B.R. and R. Pradel (2004). Estimating survival and temporary emigration in the multistate capture-recapture framework. Ecology 85: 2107-2113. Codes.\nAncrenaz M., Goossens B., Gimenez O., Sawang A. and I. Lackman-Ancrenaz (2004). Determination of ape distribution and population size using ground and aerial surveys: a case study with orang-utans in lower Kinabatangan, Sabah, Malaysia. Animal Conservation 7: 375-385.\nCharpentier M., Hossaert-McKey M., Peignot P., Gimenez O., Setchell J.M. and E.J. Wickings (2004). Social factors and male reproductive success in a highly sexually dimorphic primate: Paternity analysis over a 20 year period in a group of mandrills (Mandrillus sphinx). Folia Primatologica 75: 244-245.\nGimenez O., Choquet R. and J.-D. Lebreton (2003). Parameter redundancy in multistate capture-recapture models. Biometrical Journal 45: 704-722. Codes.\nPradel R., Wintrebert C.M.A. and Gimenez O. (2003). A proposal for a goodness-of-fit test to the Arnason-Schwarz multisite capture-recapture model. Biometrics 59: 43-53.\nBesnard A., Gimenez O. and J.-D. Lebreton (2002). A model for the evolution of creching behaviour in gulls. Evolutionary Ecology 16: 489-503.\n","date":1697846400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1697846400,"objectID":"79a6ff5fd591560441afe05143df5626","permalink":"https://oliviergimenez.github.io/publication/papers/","publishdate":"2023-10-21T00:00:00Z","relpermalink":"/publication/papers/","section":"publication","summary":"Preprints\nMoulherat S., L. Pautrel, G. Debat, M.-P. Etienne, L. Gendron, N. Hautière, J.-P. Tarel, G. Testud, O. Gimenez (2023). Biodiversity Monitoring with Intelligent Sensors: An Integrated Pipeline for Mitigating Animal-Vehicle Collisions.","tags":null,"title":"Papers","type":"publication"},{"authors":null,"categories":null,"content":"July 12-13, 2023. Workshop on Bayesian capture-recapture inference with hidden Markov models in R and Nimble for the València International Bayesian Analysis Summer School. Slides, worksheets and video recordings freely available from dedicated website.\nJune 1-2, 2023. Quantitative methods for population dynamics: A two-day workshop with R. All material available from dedicated website.\nMay 16, 2023. Workshop on estimating (wolf) occupancy with R for the Wolves Across Borders conference. All material available from dedicated website.\nJune 26, 2022. Workshop on Bayesian capture-recapture inference with hidden Markov models, R and Nimble for the ISEC2022 conference. All material available from dedicated website.\nMarch 21-22, 2022. Quantitative methods for population dynamics: A two-day workshop with R. All material available from dedicated website.\nMarch 19, 2022. Workshop on NIMBLE. All material available from dedicated website.\nJune 21-22, 2021. Workshop on individual-based models with R. All material available from dedicated website.\nMay 17-18, 2021. Workshop on Bayesian capture-recapture inference with hidden Markov models in R and Nimble. Slides, worksheets and video recordings freely available from dedicated website.\nMarch 22-26, 2021. Bayesian statistics with R for non-stats PhD students. Slides and videos here.\nMarch 18, 2021. Workshop on reproducible science in R and RStudio. Remotely via Zoom. Slides and videos here.\nNovember 23-27, 2020. Integrated population models. Remotely via Zoom. Slides and videos on capture-recapture models here.\nNovember 2020. Bonnes pratiques pour une recherche reproductible en écologie numérique. Montpellier, France.\nOctober 2020. Statistics for ecologists Master course.\nJuly 2020. Bayesian statistics with R and Jags. Remotely via Zoom.\nDécembre 2019. Bonnes pratiques pour une recherche reproductible en écologie numérique. Montpellier, France.\n2010-2018. Multievent capture-recapture models. Montpellier, France.\n","date":1689984000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1689984000,"objectID":"4139306fa16fab1c8cb885cd90f0622d","permalink":"https://oliviergimenez.github.io/talks/workshop/","publishdate":"2023-07-22T00:00:00Z","relpermalink":"/talks/workshop/","section":"talks","summary":"July 12-13, 2023. Workshop on Bayesian capture-recapture inference with hidden Markov models in R and Nimble for the València International Bayesian Analysis Summer School. Slides, worksheets and video recordings freely available from dedicated website.","tags":null,"title":"Workshops","type":"talks"},{"authors":null,"categories":null,"content":" Peyrard N. and Gimenez O. (2022). Statistical Approaches for Hidden Variables in Ecology. La version française est ici. Newman, K.B., S.T. Buckland, B.J.T. Morgan, R. King, D.L. Borchers, D.J. Cole, P. Besbeas, O. Gimenez and L. Thomas (2014). Modelling Population Dynamics - Model Formulation, Fitting and Assessment using State-Space Methods. King, R., B.J.T. Morgan, O. Gimenez, S.P. Brooks (2009). Bayesian Analysis of Population Ecology. Morrissey, M.B., P. de Villemereuil, B. Doligez, O. Gimenez (2014). Bayesian approaches to the quantitative genetic analysis of natural populations. In Quantitative Genetics in the Wild, Editors: A. Charmantier, D. Garant, and L.E.B. Kruuk. Oxford University Press, pages 228-253. Grosbois V. and O. Gimenez (2010). Capture-mark-recapture models. In Effects of Climate Change on Birds, Editors: A.P. Moller, W. Fiedler and P. Berthold, Oxford University Press, pages 39-46. Lebreton, J.-D., E. Klein, O. Gimenez and F. Rousset (2010). Statistiques pour la biologie évolutive. Dans Biologie Evolutive, Editeurs: Thomas F, Lefèvre T \u0026amp; Raymond M. De Boeck Editions, pages 33-48. Rouan L., S. Cubaynes, C. Duchamp, C. Miquel, A.-.M. Reboulet, O. Gimenez, J.-D. Lebreton, R. Choquet and R. Pradel (2010). Modèles de mélange en capture-recapture. Dans Modèles à variables latentes et modèles de mélange, Editeurs: G. Saporta, J.J. Droesbeke and C. Thomas-Agnan, Editions Technip, pages 265-272. ","date":1666569600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1666569600,"objectID":"e6b9c30b973c4d01c41005b0bd7d4026","permalink":"https://oliviergimenez.github.io/publication/books/","publishdate":"2022-10-24T00:00:00Z","relpermalink":"/publication/books/","section":"publication","summary":"Peyrard N. and Gimenez O. (2022). Statistical Approaches for Hidden Variables in Ecology. La version française est ici. Newman, K.B., S.T. Buckland, B.J.T. Morgan, R. King, D.L. Borchers, D.J. Cole, P.","tags":null,"title":"Books and book chapters","type":"publication"},{"authors":null,"categories":null,"content":"Un point d\u0026#39;étape sur nos projets sur la loutre d\u0026#39;Europe. J\u0026#39;y parle de connectivité, trame turquoise, pièges photos, ADN env, cacas, représentations des usagers et médiation scientifique. Et de caca, j\u0026#39;ai déjà dit caca? 🦦💩🧬📸\nPhotos : Yann Raulethttps://t.co/5CakMhkggi pic.twitter.com/pkhaVQwPOV\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) December 9, 2023 ","date":1702080000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1702080000,"objectID":"0c5ca14563a824b30703979501f3515f","permalink":"https://oliviergimenez.github.io/blog/updateotterconnect/","publishdate":"2023-12-09T00:00:00Z","relpermalink":"/blog/updateotterconnect/","section":"blog","summary":"Un point d\u0026#39;étape sur nos projets sur la loutre d\u0026#39;Europe. J\u0026#39;y parle de connectivité, trame turquoise, pièges photos, ADN env, cacas, représentations des usagers et médiation scientifique. Et de caca, j\u0026#39;ai déjà dit caca?","tags":["statistical ecology","Otter","scientific outreach"],"title":"Update on our Otter projects","type":"blog"},{"authors":null,"categories":null,"content":"Petit exposé sur les modèles et leurs applications en écologie à l\u0026#39;occasion d\u0026#39;un atelier d\u0026#39;échanges Modélisations « Biodiversité et Infrastructures Linéaires de Transport » organisé par @CeremaCom @OFBiodiversite @CentreTVB @Ecologie_Gouv #ITTECOP https://t.co/tLOWezfc69 pic.twitter.com/QjlOelYSMU\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) November 28, 2023 ","date":1701129600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1701129600,"objectID":"19fc9e1393fe13c1f29bff2316b519c7","permalink":"https://oliviergimenez.github.io/blog/modelscerema/","publishdate":"2023-11-28T00:00:00Z","relpermalink":"/blog/modelscerema/","section":"blog","summary":"Petit exposé sur les modèles et leurs applications en écologie à l\u0026#39;occasion d\u0026#39;un atelier d\u0026#39;échanges Modélisations « Biodiversité et Infrastructures Linéaires de Transport » organisé par @CeremaCom @OFBiodiversite @CentreTVB @Ecologie_Gouv #ITTECOP https://t.","tags":["statistical ecology","models","wolf"],"title":"Models and applications in ecology","type":"blog"},{"authors":null,"categories":null,"content":"La team loutre est sur place ! Venez vous balader au festival des sciences participatives et citoyennes https://t.co/N6I94VU2iJ#VaSavoir @instantscience @cefemontpellier @CNRSecologie @CNRS_OccitaniE @montpellier_ 🦦https://t.co/CHEJnrW9KE pic.twitter.com/tSe2uSnHsG\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) October 28, 2023 ","date":1698451200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1698451200,"objectID":"2bc6b33a1b6a004c08ed3260b5ace586","permalink":"https://oliviergimenez.github.io/blog/instantscience/","publishdate":"2023-10-28T00:00:00Z","relpermalink":"/blog/instantscience/","section":"blog","summary":"La team loutre est sur place ! Venez vous balader au festival des sciences participatives et citoyennes https://t.co/N6I94VU2iJ#VaSavoir @instantscience @cefemontpellier @CNRSecologie @CNRS_OccitaniE @montpellier_ 🦦https://t.co/CHEJnrW9KE pic.twitter.com/tSe2uSnHsG\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) October 28, 2023 ","tags":["scientific outreach","citizen science","eurasian otters"],"title":"Otters and citizen science","type":"blog"},{"authors":null,"categories":null,"content":"Excited to be running a workshop on Bayesian capture-recapture inference w hidden Markov models, #rstats \u0026amp; @R_nimble 🥳\nThanks for the invitation @vibass7 🥰\nWebsite is up and running w slides, code \u0026amp; data https://t.co/X7MYs8ec5X pic.twitter.com/UDzIVeVhuZ\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) July 10, 2023 ","date":1688947200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1688947200,"objectID":"b861076dda9838d52138ed8ae9cbda13","permalink":"https://oliviergimenez.github.io/blog/vibassworkshop/","publishdate":"2023-07-10T00:00:00Z","relpermalink":"/blog/vibassworkshop/","section":"blog","summary":"Excited to be running a workshop on Bayesian capture-recapture inference w hidden Markov models, #rstats \u0026amp; @R_nimble 🥳\nThanks for the invitation @vibass7 🥰\nWebsite is up and running w slides, code \u0026amp; data https://t.","tags":["workshop","capture-recapture","nimble","bayes","rstats"],"title":"Workshop on Bayesian analysis of capture-recapture data with hidden Markov models","type":"blog"},{"authors":null,"categories":null,"content":"Au festival #AgropolEat une sortie organisée avec mon collègue Vincent Sablain #EPTBLez pour sensibiliser à la présence de la loutre d\u0026#39;Europe sur le Lez 🦦💩📸🧬\nBonne nouvelle pour la biodiversité, de nouveaux équilibres à construire\nInfos 👉 https://t.co/CHEJnrW9KE\nVidéos ⤵️ pic.twitter.com/E1nwD2xRJM\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) July 2, 2023 ","date":1688256000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1688256000,"objectID":"80175ccb86ece5813d407b12d613f3fa","permalink":"https://oliviergimenez.github.io/blog/agropoleat/","publishdate":"2023-07-02T00:00:00Z","relpermalink":"/blog/agropoleat/","section":"blog","summary":"Au festival #AgropolEat une sortie organisée avec mon collègue Vincent Sablain #EPTBLez pour sensibiliser à la présence de la loutre d\u0026#39;Europe sur le Lez 🦦💩📸🧬\nBonne nouvelle pour la biodiversité, de nouveaux équilibres à construire","tags":["festival","OtterConnect","vulgarisation"],"title":"La loutre au festival Agropol'Eat","type":"blog"},{"authors":null,"categories":null,"content":"Notre labo @cefemontpellier a eu la visite du PDG @CNRS \u0026amp; du président @umontpellier, à qui j\u0026#39;ai présenté nos travaux sur les gds carnivores \u0026amp; l\u0026#39;écologie statistique 🐺🔍🧮@antoine_petit_ @Philippe_Auge @NavasLaure @jerome_vitre @OFBiodiversite 🧵▶️ https://t.co/IIV3aX2nKo pic.twitter.com/JR252ix8rV\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) June 20, 2023 ","date":1687219200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1687219200,"objectID":"cf450f6d85de55227be42c422d106b17","permalink":"https://oliviergimenez.github.io/blog/visitepdg/","publishdate":"2023-06-20T00:00:00Z","relpermalink":"/blog/visitepdg/","section":"blog","summary":"Notre labo @cefemontpellier a eu la visite du PDG @CNRS \u0026amp; du président @umontpellier, à qui j\u0026#39;ai présenté nos travaux sur les gds carnivores \u0026amp; l\u0026#39;écologie statistique 🐺🔍🧮@antoine_petit_ @Philippe_Auge @NavasLaure @jerome_vitre @OFBiodiversite 🧵▶️ https://t.","tags":["CNRS"],"title":"Visite du PDG du CNRS au CEFE","type":"blog"},{"authors":null,"categories":null,"content":"Really proud of this new paper in @AER_ESE_BES in which we give our two cents on how to fill the gap bw researchers \u0026amp; managers w/ our long-term collab bw our lab @cefemontpellier \u0026amp; the French Biodiversity Agency @OFBiodiversite https://t.co/JrMcpFr9yl@CNRS_OccitaniE 🧵👇 pic.twitter.com/D7RO6VZA1W\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) June 7, 2023 ","date":1686096000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1686096000,"objectID":"b000ee28e5ece73efb033542d3a05ad5","permalink":"https://oliviergimenez.github.io/blog/newpaperofb/","publishdate":"2023-06-07T00:00:00Z","relpermalink":"/blog/newpaperofb/","section":"blog","summary":"Really proud of this new paper in @AER_ESE_BES in which we give our two cents on how to fill the gap bw researchers \u0026amp; managers w/ our long-term collab bw our lab @cefemontpellier \u0026amp; the French Biodiversity Agency @OFBiodiversite https://t.","tags":["research management gap","OFB","CNRS"],"title":"New paper alert","type":"blog"},{"authors":null,"categories":null,"content":"Pour Occitanie Est, la semaine \u0026quot;écologie environnement biodiversité\u0026quot; aura lieu le 13 juin 14h30 et on y parlera des travaux des jeunes chercheurs et chercheuses, et des collaborations avec la ville/métro de Montpellier. Inscriptions jusqu\u0026#39;à vendredi ▶️ https://t.co/6Hw2BAvGAV https://t.co/mWwGICvaqZ pic.twitter.com/4ryMBcnhKL\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) June 7, 2023 ","date":1686096000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1686096000,"objectID":"c835359b48ec765b923533c1d37aaad3","permalink":"https://oliviergimenez.github.io/blog/semaineinee2023/","publishdate":"2023-06-07T00:00:00Z","relpermalink":"/blog/semaineinee2023/","section":"blog","summary":"Pour Occitanie Est, la semaine \u0026quot;écologie environnement biodiversité\u0026quot; aura lieu le 13 juin 14h30 et on y parlera des travaux des jeunes chercheurs et chercheuses, et des collaborations avec la ville/métro de Montpellier.","tags":["conférence","biodiversité","CNRS","INEE"],"title":"Semaine écologie environnement biodiversité du CNRS","type":"blog"},{"authors":null,"categories":null,"content":"🥳🤩😇 yet another great paper by @ValentinLauret from his PhD w/ Hélène Labach, Léa David and Matthieu Authier 😍\nthx @masonfidino @PaulCaplat 😉@cefemontpellier @CNRS_OccitaniE @INEE_CNRS @MIRACETI_ceta @UnivLaRochelle @umontpellier https://t.co/GT7pdMXauN\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) June 2, 2023 ","date":1685664000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1685664000,"objectID":"6cf0cbbf820d2feaf73a83080887db99","permalink":"https://oliviergimenez.github.io/blog/newpaperval/","publishdate":"2023-06-02T00:00:00Z","relpermalink":"/blog/newpaperval/","section":"blog","summary":"🥳🤩😇 yet another great paper by @ValentinLauret from his PhD w/ Hélène Labach, Léa David and Matthieu Authier 😍\nthx @masonfidino @PaulCaplat 😉@cefemontpellier @CNRS_OccitaniE @INEE_CNRS @MIRACETI_ceta @UnivLaRochelle @umontpellier https://t.co/GT7pdMXauN\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) June 2, 2023 ","tags":["multispecies occupancy","bottlenose dolphins","photo-id"],"title":"New paper alert","type":"blog"},{"authors":null,"categories":null,"content":"Aaaaaand let\u0026#39;s start this thing 😉 @abesnardEPHE @SarahCubaynes @javi_ferlop @ValentinLauret @MarwanNaciri2 @ThibautCouturie @LiseViollat @cthierrycc\n▶️ https://t.co/jlTE4inuHM@cefemontpellier @INEE_CNRS @CNRS_OccitaniE @EPHE_PSL @umontpellier https://t.co/Iehm6DdhjM\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) June 1, 2023 ","date":1685577600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1685577600,"objectID":"4befbd8d289aca8ecb979b9cd7dea38d","permalink":"https://oliviergimenez.github.io/blog/workshoppopdyn2023/","publishdate":"2023-06-01T00:00:00Z","relpermalink":"/blog/workshoppopdyn2023/","section":"blog","summary":"Aaaaaand let\u0026#39;s start this thing 😉 @abesnardEPHE @SarahCubaynes @javi_ferlop @ValentinLauret @MarwanNaciri2 @ThibautCouturie @LiseViollat @cthierrycc\n▶️ https://t.co/jlTE4inuHM@cefemontpellier @INEE_CNRS @CNRS_OccitaniE @EPHE_PSL @umontpellier https://t.co/Iehm6DdhjM\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) June 1, 2023 ","tags":["workshop","population dynamics","rstats","team work"],"title":"Workshop on population dynamics","type":"blog"},{"authors":null,"categories":null,"content":"It\u0026#39;s #WorldOtterDay and we\u0026#39;re happy to share our research on the Eurasian otter and socio-ecological connectivity using cam traps, eDNA, spraints, interviews #interdisciplinarity\nCheck out https://t.co/QwwJtY2Aeq for info @IOSF @montpellier_ @Montpellier3m @umontpellier 🎥👇 pic.twitter.com/dfphlFdwY6\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) May 31, 2023 ","date":1685491200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1685491200,"objectID":"4cde6f12900e63743f6e69e410b082dc","permalink":"https://oliviergimenez.github.io/blog/worldotterday/","publishdate":"2023-05-31T00:00:00Z","relpermalink":"/blog/worldotterday/","section":"blog","summary":"It\u0026#39;s #WorldOtterDay and we\u0026#39;re happy to share our research on the Eurasian otter and socio-ecological connectivity using cam traps, eDNA, spraints, interviews #interdisciplinarity\nCheck out https://t.co/QwwJtY2Aeq for info @IOSF @montpellier_ @Montpellier3m @umontpellier 🎥👇 pic.","tags":["eurasian otter","OtterConnect"],"title":"World Otter Day","type":"blog"},{"authors":null,"categories":null,"content":"🚨 New paper 🚨 by @PatriRodrigs testing the \u0026quot;humans as shield\u0026quot; \u0026amp; \u0026quot;humans as super-predators\u0026quot; hypotheses w/ camp trap data on mammals using multispecies occupancy models à la Rota et al. 🤩😇https://t.co/NMlm3loxG8 pic.twitter.com/P4upBAzb8g\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) May 30, 2023 ","date":1685404800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1685404800,"objectID":"688eacab394dc33981f65d3fd8531b67","permalink":"https://oliviergimenez.github.io/blog/newpaperpat/","publishdate":"2023-05-30T00:00:00Z","relpermalink":"/blog/newpaperpat/","section":"blog","summary":"🚨 New paper 🚨 by @PatriRodrigs testing the \u0026quot;humans as shield\u0026quot; \u0026amp; \u0026quot;humans as super-predators\u0026quot; hypotheses w/ camp trap data on mammals using multispecies occupancy models à la Rota et al.","tags":["multispecies occupancy","human shield hypothesis","camera traps"],"title":"New paper alert","type":"blog"},{"authors":null,"categories":null,"content":"New preprint led by Marine Desprez w/ Vincent Miele on \u0026quot;Nine tips for ecologists using machine learning\u0026quot; 🤩#MachineLearning #Ecology #RStats #preprinthttps://t.co/KvXTmhABtF@LbbeLyon @cefemontpellier @INEE_CNRS @CNRS_OccitaniE @umontpellier pic.twitter.com/CbiFOIjXhX\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) May 22, 2023 ","date":1684713600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1684713600,"objectID":"238aecd74817a3e86678b9a9ed7ba7ce","permalink":"https://oliviergimenez.github.io/blog/preprintml/","publishdate":"2023-05-22T00:00:00Z","relpermalink":"/blog/preprintml/","section":"blog","summary":"New preprint led by Marine Desprez w/ Vincent Miele on \u0026quot;Nine tips for ecologists using machine learning\u0026quot; 🤩#MachineLearning #Ecology #RStats #preprinthttps://t.co/KvXTmhABtF@LbbeLyon @cefemontpellier @INEE_CNRS @CNRS_OccitaniE @umontpellier pic.twitter.com/CbiFOIjXhX\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) May 22, 2023 ","tags":["machine learning","computer vision","tips","deep learning","artificial intelligence","rstats"],"title":"Nine tips for ecologists using machine learning","type":"blog"},{"authors":null,"categories":null,"content":"I just run a 3h workshop on estimating occupancy in R w/ wolf as a case study 🐺🗺️ 🌍 Website w/ slides, R code and data https://t.co/KOKaPRPRtv\n📽️ Video recording (in English w/ 👽 accent) https://t.co/LlmkLtrd8L#WolvesBorders @WolvesBorders #RStats 🧵👇 pic.twitter.com/yz8gb0Hc3C\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) May 16, 2023 ","date":1684195200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1684195200,"objectID":"5af3a6e4d07ac236b808779dd6226dc7","permalink":"https://oliviergimenez.github.io/blog/workshopoccupancy/","publishdate":"2023-05-16T00:00:00Z","relpermalink":"/blog/workshopoccupancy/","section":"blog","summary":"I just run a 3h workshop on estimating occupancy in R w/ wolf as a case study 🐺🗺️ 🌍 Website w/ slides, R code and data https://t.co/KOKaPRPRtv\n📽️ Video recording (in English w/ 👽 accent) https://t.","tags":["statistical ecology","occupancy","rstats","unmarked"],"title":"Workshop on estimating occupancy in R","type":"blog"},{"authors":null,"categories":null,"content":"Le Réseau Thématique en écologie statistique propose un annuaire des labos/pers en #ecostat Que vous soyez à la recherche d\u0026#39;un labo pour un stage, d\u0026#39;une pers avec une compétence, ou pour vous faire une idée des contours de notre communauté, visitez https://t.co/jaXdqPFVKe 🤩🥳 pic.twitter.com/qSrS7XHrso\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) May 5, 2023 ","date":1683244800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1683244800,"objectID":"d743300396898aa9d5e1125d253e9003","permalink":"https://oliviergimenez.github.io/blog/networkgdrecostat/","publishdate":"2023-05-05T00:00:00Z","relpermalink":"/blog/networkgdrecostat/","section":"blog","summary":"Le Réseau Thématique en écologie statistique propose un annuaire des labos/pers en #ecostat Que vous soyez à la recherche d\u0026#39;un labo pour un stage, d\u0026#39;une pers avec une compétence, ou pour vous faire une idée des contours de notre communauté, visitez https://t.","tags":["statistical ecology","network"],"title":"Annuaire du Réseau Thématique en écologie statistique","type":"blog"},{"authors":null,"categories":["project"],"content":"Context. Conservation and the sustainable use of biodiversity and ecosystem services largely depend on the successful management of ecological populations, which in turn depend on our knowledge and our ability to understand and predict how the structure and dynamics of ecological populations respond to changes in their environments.\nRationale and aim. Despite its importance, the quantification of the effects of anthropic pressures on biodiversity is difficult and rarely undertaken because of the lack of species-specific ecological data and/or human pressures and activities, and the complexity and lack of transfer of analytical methods needed to evaluate these long-term consequences. The DISCAR (DIS = disturbance, CAR = carnivores) project aims at providing an operational framework with analytical tools for assessing the impacts or population consequences of human pressures, and showcasing it with case studies in applied conservation.\nCase study. In DISCAR, we will use small carnivores in French mainland and overseas territories as a case study to assess the population consequences or impacts of human pressures on animal populations. Why small carnivores? Because these species hold important roles in ecosystems, such as influencing ecosystem structure and providing numerous ecosystem services, including pest/disease control and seed dispersal. Small carnivore species also respond to ecosystem perturbation in a fast, measurable, and interpretable way and therefore can be used as sentinels of global change.\nObjectives. DISCAR will connect the fields of bioenergetics and demographic modeling in an integrated framework to assess human pressures on biodiversity, and to characterize then predict the impacts on small carnivores. DISCAR will identify key parameters underlying coupling between pressures and impacts, and will make recommendations for stakeholders in terms of monitoring and mitigation strategies.\nWork packages. We will i) assemble a unique dataset on small carnivores including their ecology, bioenergetic requirements, demography and human pressures, ii) quantify and characterize the impacts of interacting pressures on populations, and iii) forecast the fate of populations to assess their viability under various scenarios of pressures and mitigation to help devise sound conservation strategies.\nConsortium. To achieve these objectives, we have gathered international experts in ecological modelling and small carnivores who will develop and transfer an analytical framework for quantification, characterization and prediction of the impacts of pressures on populations. These experts are:\nOlivier Gimenez – CNRS (France) Sandrine Ruette – OFB (France) Courtney Marneweck – Giraffe Conservation Foundation (South Africa) Patrick Giraudoux – Université Bourgogne Franche-Comté (France) Isabel Smallegange - Newcastle University (UK) Sébastien Devillard – Université Claude Bernard Lyon 1 (France) Cara Gallagher - University of Potsdam (Allemagne) Sarah Cubaynes – EPHE (France) John Fryxell – University of Guelph (Canada) Elizabeth McHuron – University of Washington (USA) Rob Salguero-Gomez – University of Oxford (UK) Marie-Lazarine Poulle – Université Reims Champagne Ardenne (France) Funding. This project is funded by the French Foundation for Biodiversity Research (FRB) through Cesab.\n","date":1683072000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1683072000,"objectID":"40cfee82f289cf02f742bd4383ac7165","permalink":"https://oliviergimenez.github.io/my-project/external-project/discar/","publishdate":"2023-05-03T00:00:00Z","relpermalink":"/my-project/external-project/discar/","section":"my-project","summary":"Population consequences of human DISturbance on small CARnivores.","tags":["small carnivores","monitoring","management"],"title":"DISCAR","type":"my-project"},{"authors":null,"categories":["project"],"content":"","date":1683072000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1683072000,"objectID":"1a1d94cebff1799227ca00b4ec8a8395","permalink":"https://oliviergimenez.github.io/my-project/external-project/otters/","publishdate":"2023-05-03T00:00:00Z","relpermalink":"/my-project/external-project/otters/","section":"my-project","summary":"An applied research project to better understand otter recolonization.","tags":["carnivores","monitoring","eDNA","cam-traps"],"title":"OtterConnect","type":"my-project"},{"authors":null,"categories":null,"content":"Last week we had the pleasure to host the #euring2023 conf 🤩 Awesome pple, amazing science, sun \u0026amp; good food 😇 Hope everyone enjoyed it 💜\nVideo recordings of talks \u0026amp; workshops freely available at https://t.co/aPG6PXHw6Q PDF at https://t.co/i3Ew3sbe9h\n🧵👇 w/ many 🙏 1/15\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) April 24, 2023 ","date":1682294400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1682294400,"objectID":"5ec6e01115fffa541132ce28f774c719","permalink":"https://oliviergimenez.github.io/blog/euring2023/","publishdate":"2023-04-24T00:00:00Z","relpermalink":"/blog/euring2023/","section":"blog","summary":"Last week we had the pleasure to host the #euring2023 conf 🤩 Awesome pple, amazing science, sun \u0026amp; good food 😇 Hope everyone enjoyed it 💜\nVideo recordings of talks \u0026amp; workshops freely available at https://t.","tags":["statistical ecology","euring","conference"],"title":"Euring 2023","type":"blog"},{"authors":null,"categories":null,"content":"Jetez un coup d\u0026rsquo;oeil à https://youtu.be/xbUoXrEsDV4.\n😇🥳 Super reportage @F3OccitanieLR sur notre projet loutre OtterConnect https://t.co/CHEJnrW9KE avec Vincent Sablain #EPTBLez 🦦💩📸🧬 Soutenu par @montpellier_ @Montpellier3m @umontpellier @zoobeauval @cefemontpellier ▶️ 8\u0026#39;36\u0026#39;\u0026#39; https://t.co/JFp8dNJhxc\n🙏 @SebastienBanus\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) April 24, 2023 ","date":1682294400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1682294400,"objectID":"ee4dacdb28664d2d248030001cc11e66","permalink":"https://oliviergimenez.github.io/blog/reportagefrance3/","publishdate":"2023-04-24T00:00:00Z","relpermalink":"/blog/reportagefrance3/","section":"blog","summary":"Jetez un coup d\u0026rsquo;oeil à https://youtu.be/xbUoXrEsDV4.\n😇🥳 Super reportage @F3OccitanieLR sur notre projet loutre OtterConnect https://t.co/CHEJnrW9KE avec Vincent Sablain #EPTBLez 🦦💩📸🧬 Soutenu par @montpellier_ @Montpellier3m @umontpellier @zoobeauval @cefemontpellier ▶️ 8\u0026#39;36\u0026#39;\u0026#39; https://t.","tags":["otter","media","reportage","otterconnect"],"title":"Reportage France 3 sur notre projet loutre OtterConnect","type":"blog"},{"authors":null,"categories":null,"content":"Oksana Grente @oksanagrente a fait sa thèse sur les déprédations par le loup et le contrôle létal 🐺🐑 Pour porter ses résultats à connaissance de toutes et tous, elle a écrit cet article https://t.co/UaBgeWdySu. C\u0026#39;est en français, et plus court que le manuscrit de \u0026gt; 300 pages 😉 https://t.co/HgeoLlYe3P pic.twitter.com/tNDw8pIEzm\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) April 6, 2023 ","date":1680739200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1680739200,"objectID":"d8751db68a97561b6e725d9fe0cf6962","permalink":"https://oliviergimenez.github.io/blog/naturae/","publishdate":"2023-04-06T00:00:00Z","relpermalink":"/blog/naturae/","section":"blog","summary":"Oksana Grente @oksanagrente a fait sa thèse sur les déprédations par le loup et le contrôle létal 🐺🐑 Pour porter ses résultats à connaissance de toutes et tous, elle a écrit cet article https://t.","tags":["wildlife management","wolf","paper"],"title":"déprédations par le loup et contrôle létal","type":"blog"},{"authors":null,"categories":null,"content":"I\u0026rsquo;m starting a new project on the Eurasian otter to better understand its recolonization in France.\nIn the OtterConnect project, we aim at providing an operational framework for assessing socio-ecological connectivity for small carnivores which influence the dynamics of ecosystems and provide numerous ecosystem services. We focus on the Eurasian otter (Lutra lutra) an umbrella and indicator species of blue and green infrastructures. To know more about the project and get news, please visit our website https://otterconnect.netlify.app/en/.\n","date":1680480000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1680480000,"objectID":"1aa8b6063edf649f82c39891754ce034","permalink":"https://oliviergimenez.github.io/blog/otterconnect/","publishdate":"2023-04-03T00:00:00Z","relpermalink":"/blog/otterconnect/","section":"blog","summary":"I\u0026rsquo;m starting a new project on the Eurasian otter to better understand its recolonization in France.\n","tags":["small carnivores","eDNA","cam-traps","monitoring"],"title":"A new project on the Eurasian otter and connectivity","type":"blog"},{"authors":null,"categories":null,"content":"How to assess landscape connectivity with spatial capture-recapture models? @oSCR_package to the rescue ! Some code to reproduce results from a great paper by Dana Morin @Fuller_Lab @andyroyle_pwrc \u0026amp; @chrissuthy Post: https://t.co/FVwBzMnqA9\nCode: https://t.co/FMMyF0s22x pic.twitter.com/QUlOjlGnPf\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) March 27, 2023 ","date":1679875200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1679875200,"objectID":"19aa9591e6698352b661513d1a604187","permalink":"https://oliviergimenez.github.io/blog/connectoscr/","publishdate":"2023-03-27T00:00:00Z","relpermalink":"/blog/connectoscr/","section":"blog","summary":"How to assess landscape connectivity with spatial capture-recapture models? @oSCR_package to the rescue ! Some code to reproduce results from a great paper by Dana Morin @Fuller_Lab @andyroyle_pwrc \u0026amp; @chrissuthy Post: https://t.","tags":["connectivity","R","rstats"],"title":"How to assess landscape connectivity with spatial capture-recapture models?","type":"blog"},{"authors":null,"categories":null,"content":"📢👋 Together with @SarahCubaynes \u0026amp; @abesnardEPHE we will give a 2-day introductory workshop June 1-2 on quantitative methods for population dynamics in R #rstats Join us 🥳 It\u0026#39;s online and free of charge 😉 You just need to register via https://t.co/otguiSMTY0\nPlease RT 😇 pic.twitter.com/6ghDHE92SJ\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) February 8, 2023 ","date":1675814400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1675814400,"objectID":"8eefd2814c017810c2eb7201267c404a","permalink":"https://oliviergimenez.github.io/blog/annonceworkshoppopdyn/","publishdate":"2023-02-08T00:00:00Z","relpermalink":"/blog/annonceworkshoppopdyn/","section":"blog","summary":"📢👋 Together with @SarahCubaynes \u0026amp; @abesnardEPHE we will give a 2-day introductory workshop June 1-2 on quantitative methods for population dynamics in R #rstats Join us 🥳 It\u0026#39;s online and free of charge 😉 You just need to register via https://t.","tags":["population dynamics","workshop","R","rstats"],"title":"Workshop on quantitative methods for population dynamics in R","type":"blog"},{"authors":null,"categories":null,"content":"Ey! VIBASS 6 is announced online now https://t.co/VS8cLR8ocK\n🗓️ July 10-14 2023\n🚩Faculty of Mathematics @FMatemaUV_EG, Valencia, Spain Be aware that the registration would start really really soon and we just have few spots. Don\u0026#39;t miss it. pic.twitter.com/5M61dbuxuF\n\u0026mdash; VIBASS 6 (@vibass7) January 13, 2023 ","date":1673827200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1673827200,"objectID":"6618d54253a8831cb977ce9bb200d9db","permalink":"https://oliviergimenez.github.io/blog/annoncevibass/","publishdate":"2023-01-16T00:00:00Z","relpermalink":"/blog/annoncevibass/","section":"blog","summary":"Ey! VIBASS 6 is announced online now https://t.co/VS8cLR8ocK\n🗓️ July 10-14 2023\n🚩Faculty of Mathematics @FMatemaUV_EG, Valencia, Spain Be aware that the registration would start really really soon and we just have few spots.","tags":["capture-recapture","hmm","workshop","bayesian statistics","R","rstats"],"title":"VIBASS and Bayesian capture-recapture workshop","type":"blog"},{"authors":null,"categories":null,"content":"Check it out, this is a cool ++++ paper by @MaelisKervellec 🤩🤩 So glad for her that she got her first PhD paper published #OpenAccess @OFBiodiversite @CNRS_OccitaniE @INEE_CNRS @umontpellier @twitthair1 @cefemontpellier https://t.co/sVdXl40DyU\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) January 11, 2023 ","date":1673395200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1673395200,"objectID":"36368e03349387c2981d323096bc747b","permalink":"https://oliviergimenez.github.io/blog/papiermaelis/","publishdate":"2023-01-11T00:00:00Z","relpermalink":"/blog/papiermaelis/","section":"blog","summary":"Check it out, this is a cool ++++ paper by @MaelisKervellec 🤩🤩 So glad for her that she got her first PhD paper published #OpenAccess @OFBiodiversite @CNRS_OccitaniE @INEE_CNRS @umontpellier @twitthair1 @cefemontpellier https://t.","tags":["spatial capture-recapture","connectivity","brown bear"],"title":"Maëlis Kervellec's first PhD paper!","type":"blog"},{"authors":null,"categories":null,"content":"This morning we\u0026#39;ll have a panorama of the math/stat/info models we use/develop at @cefemontpellier to address questions in ecology \u0026amp; evolution 🤩😇 Program below. This is an internal seminar to get to know each other better, but @Master_RAINET students will be live tweeting 😍🙏 https://t.co/qp6qYMGlOh pic.twitter.com/xrSXIIBZhm\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) December 6, 2022 ","date":1670284800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1670284800,"objectID":"ea89b7fba946012f6c8d9a2b647bd4bc","permalink":"https://oliviergimenez.github.io/blog/modelocefe/","publishdate":"2022-12-06T00:00:00Z","relpermalink":"/blog/modelocefe/","section":"blog","summary":"This morning we\u0026#39;ll have a panorama of the math/stat/info models we use/develop at @cefemontpellier to address questions in ecology \u0026amp; evolution 🤩😇 Program below. This is an internal seminar to get to know each other better, but @Master_RAINET students will be live tweeting 😍🙏 https://t.","tags":["modelling"],"title":"Modelling in ecology and evolution at CEFE","type":"blog"},{"authors":null,"categories":null,"content":"So proud of our latest paper on estimating abundance of Brown bear in the Pyrenees 🐻📈 in @PeerComJournal led by C Vanpé 😇https://t.co/aBZtWmO1UW\nYet another example of our collaboration w/ @OFBiodiversite 🤩\nData \u0026amp; code https://t.co/uZGQ3Fy2Iq#DiamondOpenAccess pic.twitter.com/H0PfGWFFxD\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) November 21, 2022 ","date":1668988800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1668988800,"objectID":"6fab3b283002d7f043648e5164f1dbd4","permalink":"https://oliviergimenez.github.io/blog/bearsize/","publishdate":"2022-11-21T00:00:00Z","relpermalink":"/blog/bearsize/","section":"blog","summary":"So proud of our latest paper on estimating abundance of Brown bear in the Pyrenees 🐻📈 in @PeerComJournal led by C Vanpé 😇https://t.co/aBZtWmO1UW\nYet another example of our collaboration w/ @OFBiodiversite 🤩","tags":["brown bear","capture-recapture"],"title":"Brown bear abundance in the Pyrenees mountains","type":"blog"},{"authors":null,"categories":null,"content":"Conditions de travail et santé mentale des jeunes chercheur.e.s : constats et solutions. La vidéo de la conf d\u0026#39;Adèle Combes @Vies_de_these et de la table ronde qui a suivi est dispo sur https://t.co/Sv4nT7zMmE. https://t.co/A9UH0aX5Un\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) November 8, 2022 ","date":1667865600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1667865600,"objectID":"45b60aded2cbe4c118a142681c6479f9","permalink":"https://oliviergimenez.github.io/blog/adele/","publishdate":"2022-11-08T00:00:00Z","relpermalink":"/blog/adele/","section":"blog","summary":"Conditions de travail et santé mentale des jeunes chercheur.e.s : constats et solutions. La vidéo de la conf d\u0026#39;Adèle Combes @Vies_de_these et de la table ronde qui a suivi est dispo sur https://t.","tags":["mental health","academic life"],"title":"Conference on mental health and working conditions in academia","type":"blog"},{"authors":null,"categories":null,"content":".@LouvrierJulie reçoit le prix jeune chercheuse 2020 de la @sfecologie au @Salon_AdNatura #AdNatura 🥳 pour son travail de thèse sur le loup 🐺🗺 @cefemontpellier @CNRS_OccitaniE @INEE_CNRS @umontpellier @OFBiodiversite pic.twitter.com/T2dNxeuaA9\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) October 29, 2022 ","date":1667001600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1667001600,"objectID":"8e69bae4ada6ec03b0a809f906b7091c","permalink":"https://oliviergimenez.github.io/blog/juliesfe/","publishdate":"2022-10-29T00:00:00Z","relpermalink":"/blog/juliesfe/","section":"blog","summary":".@LouvrierJulie reçoit le prix jeune chercheuse 2020 de la @sfecologie au @Salon_AdNatura #AdNatura 🥳 pour son travail de thèse sur le loup 🐺🗺 @cefemontpellier @CNRS_OccitaniE @INEE_CNRS @umontpellier @OFBiodiversite pic.twitter.com/T2dNxeuaA9\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) October 29, 2022 ","tags":["PhD","French society for ecology and evolution"],"title":"Julie Louvrier reçoit le prix jeune chercheuse 2020 de la SFEE","type":"blog"},{"authors":null,"categories":null,"content":"💜💜💜 https://t.co/YWC2qSepIr\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) October 25, 2022 ","date":1666656000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1666656000,"objectID":"8e6c85c37d6704ca14c56553d0cdd2cd","permalink":"https://oliviergimenez.github.io/blog/teamlunch/","publishdate":"2022-10-25T00:00:00Z","relpermalink":"/blog/teamlunch/","section":"blog","summary":"💜💜💜 https://t.co/YWC2qSepIr\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) October 25, 2022 ","tags":["Hair team","team building"],"title":"Team lunch!","type":"blog"},{"authors":null,"categories":null,"content":"Check it out 😉🥳 https://t.co/mqyuMSm8QF\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) September 22, 2022 ","date":1663804800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1663804800,"objectID":"d3ef175cddc6396f245d4ef533ced606","permalink":"https://oliviergimenez.github.io/blog/valpaper/","publishdate":"2022-09-22T00:00:00Z","relpermalink":"/blog/valpaper/","section":"blog","summary":"Check it out 😉🥳 https://t.co/mqyuMSm8QF\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) September 22, 2022 ","tags":["integrated modelling","nimble"],"title":"Integrating distance sampling and spatial capture-recapture to assess density of bottlenose dolphins","type":"blog"},{"authors":null,"categories":null,"content":"Looking for free images or icons for your talk, website, video, etc. A few pointers that I find useful 😉https://t.co/oTioaam8Y1https://t.co/pueZRTCONFhttps://t.co/nhx3wlPHhvhttps://t.co/6XT1z2jgg5https://t.co/2KCJU3xCAPhttps://t.co/QGnYaLsSjmhttps://t.co/TVlAtmlVDz pic.twitter.com/SoUshEWAgM\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) August 18, 2022 ","date":1660780800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1660780800,"objectID":"209998c5b52eb947c81f2edf9673b5cd","permalink":"https://oliviergimenez.github.io/blog/icons/","publishdate":"2022-08-18T00:00:00Z","relpermalink":"/blog/icons/","section":"blog","summary":"Looking for free images or icons for your talk, website, video, etc. A few pointers that I find useful 😉https://t.co/oTioaam8Y1https://t.co/pueZRTCONFhttps://t.co/nhx3wlPHhvhttps://t.co/6XT1z2jgg5https://t.co/2KCJU3xCAPhttps://t.co/QGnYaLsSjmhttps://t.co/TVlAtmlVDz pic.twitter.com/SoUshEWAgM\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) August 18, 2022 ","tags":["resources","icons"],"title":"Resources on finding images and icons","type":"blog"},{"authors":null,"categories":null,"content":"How do neural networks work 🤔 What\u0026#39;s under the hood 🤯 A few pointers in the thread below with a R/statistian bias 😇 #MachineLearning #DeepLearning #NeuralNetworks #CNN #RStats 🧵⬇️ pic.twitter.com/4QF1vKoFc8\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) August 8, 2022 ","date":1659916800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1659916800,"objectID":"c3cd4ff80fdecb3816212ca467bd2b00","permalink":"https://oliviergimenez.github.io/blog/neuralnetworks/","publishdate":"2022-08-08T00:00:00Z","relpermalink":"/blog/neuralnetworks/","section":"blog","summary":"How do neural networks work 🤔 What\u0026#39;s under the hood 🤯 A few pointers in the thread below with a R/statistian bias 😇 #MachineLearning #DeepLearning #NeuralNetworks #CNN #RStats 🧵⬇️ pic.twitter.com/4QF1vKoFc8","tags":["neural networks","statistical ecology"],"title":"Resources on neural networks for ecology","type":"blog"},{"authors":null,"categories":null,"content":"I’m writing a lot over summer. Writing does not come easily to me. Experience helps of course, but I find it useful to follow some rules, like how to write an intro. Fortunately, the best writing tips are out there 😉\nMy fav 🧵👇@OpenAcademics#AcademicTwitter#AcademicChatter pic.twitter.com/v2CeZKF6jx\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) July 27, 2022 ","date":1658880000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1658880000,"objectID":"ea1fdb47e27f65f85bdfc26de7cdf4b1","permalink":"https://oliviergimenez.github.io/blog/write/","publishdate":"2022-07-27T00:00:00Z","relpermalink":"/blog/write/","section":"blog","summary":"I’m writing a lot over summer. Writing does not come easily to me. Experience helps of course, but I find it useful to follow some rules, like how to write an intro.","tags":["writing","academic life"],"title":"Writing tips","type":"blog"},{"authors":null,"categories":null,"content":"C\u0026#39;est la période de rédaction de thèse pour bcp de jeunes chercheuses/chercheurs. Une étape pas facile à passer. Un fil sur comment on s\u0026#39;y prend dans l\u0026#39;équipe. Ça vaut ce que ça vaut, c\u0026#39;est pas original, mais si ça peut aider, tant mieux.\nEt vous, quels sont vos conseils?\n🧵⬇️ pic.twitter.com/p8qw4ZoDuQ\n\u0026mdash; Olivier Gimenez 🖖🦦 (@oaggimenez) July 17, 2022 ","date":1658016000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1658016000,"objectID":"067b7c73faf0b58923c5fa0caffe97f3","permalink":"https://oliviergimenez.github.io/blog/advicephdwriting/","publishdate":"2022-07-17T00:00:00Z","relpermalink":"/blog/advicephdwriting/","section":"blog","summary":"C\u0026#39;est la période de rédaction de thèse pour bcp de jeunes chercheuses/chercheurs. Une étape pas facile à passer. Un fil sur comment on s\u0026#39;y prend dans l\u0026#39;équipe. Ça vaut ce que ça vaut, c\u0026#39;est pas original, mais si ça peut aider, tant mieux.","tags":["PhD","academic life"],"title":"Conseils pour écrire sa thèse","type":"blog"},{"authors":[],"categories":[],"content":"If you need to test the goodness-of-fit of your open capture-recapture models, good news, our #RStats R2ucare is back on CRAN thanks to the amazing work of Murray Efford and @bc_stevenson 🙏🤩\nwebsite: https://t.co/SbGrQm9XS2 paper: https://t.co/W38nqsEHPa pic.twitter.com/KelIzsB8Xf\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 12, 2022 ","date":1657670400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1657670400,"objectID":"dd07f490d66f126a7fe5df3639cd794e","permalink":"https://oliviergimenez.github.io/authors/2022-07-13-nos-travaux-sur-le-loup/","publishdate":"2022-07-13T00:00:00Z","relpermalink":"/authors/2022-07-13-nos-travaux-sur-le-loup/","section":"authors","summary":"If you need to test the goodness-of-fit of your open capture-recapture models, good news, our #RStats R2ucare is back on CRAN thanks to the amazing work of Murray Efford and @bc_stevenson 🙏🤩","tags":["loup"],"title":"Nos travaux sur le loup","type":"authors"},{"authors":[],"categories":[],"content":"Notre équipe @twitthair1 travaille bcp sur les questions en lien avec la gestion du loup en France 🐺🇫🇷 Voici qqs pointeurs sur nos travaux, en privilégiant les productions en français. Plusieurs de ces résultats ont été obtenus avec les collègues @OFBiodiversite 🧵⬇️\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 13, 2022 ","date":1657670400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1657670400,"objectID":"f7c7277e1cdf2a9d65a5558ad08a927e","permalink":"https://oliviergimenez.github.io/blog/2022-07-13-nos-travaux-sur-le-loup/","publishdate":"2022-07-13T00:00:00Z","relpermalink":"/blog/2022-07-13-nos-travaux-sur-le-loup/","section":"blog","summary":"Notre équipe @twitthair1 travaille bcp sur les questions en lien avec la gestion du loup en France 🐺🇫🇷 Voici qqs pointeurs sur nos travaux, en privilégiant les productions en français. Plusieurs de ces résultats ont été obtenus avec les collègues @OFBiodiversite 🧵⬇️","tags":["loup"],"title":"Nos travaux sur le loup","type":"blog"},{"authors":[],"categories":[],"content":"If you need to test the goodness-of-fit of your open capture-recapture models, good news, our #RStats R2ucare is back on CRAN thanks to the amazing work of Murray Efford and @bc_stevenson 🙏🤩\nwebsite: https://t.co/SbGrQm9XS2 paper: https://t.co/W38nqsEHPa pic.twitter.com/KelIzsB8Xf\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 12, 2022 ","date":1657584000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1657584000,"objectID":"294bec081990d9e0c8c03a35f410baf2","permalink":"https://oliviergimenez.github.io/blog/2022-07-12-r2ucare-back-on-cran/","publishdate":"2022-07-12T00:00:00Z","relpermalink":"/blog/2022-07-12-r2ucare-back-on-cran/","section":"blog","summary":"If you need to test the goodness-of-fit of your open capture-recapture models, good news, our #RStats R2ucare is back on CRAN thanks to the amazing work of Murray Efford and @bc_stevenson 🙏🤩","tags":["R2ucare","package","ANR"],"title":"R2ucare back on CRAN","type":"blog"},{"authors":[],"categories":[],"content":"🥳 Save the date 🥳\nThe 2023 Euring Analytical Conference will be held in Montpellier (France), on April 17-21, 2023.\nCheck out https://t.co/t04XJGbW8k for info. 🤗🤩We look forward to welcoming you all 🤗🤩@_EURING @abesnardEPHE pic.twitter.com/KAe46wOaze\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 8, 2022 ","date":1657238400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1657238400,"objectID":"b086efcbffbddc67888bc35a939c83da","permalink":"https://oliviergimenez.github.io/blog/2022-07-08-save-the-data-euring2023-in-montpellier-april-17-21/","publishdate":"2022-07-08T00:00:00Z","relpermalink":"/blog/2022-07-08-save-the-data-euring2023-in-montpellier-april-17-21/","section":"blog","summary":"🥳 Save the date 🥳\nThe 2023 Euring Analytical Conference will be held in Montpellier (France), on April 17-21, 2023.\nCheck out https://t.co/t04XJGbW8k for info. 🤗🤩We look forward to welcoming you all 🤗🤩@_EURING @abesnardEPHE pic.","tags":["euring","capture-recapture","conference"],"title":"Save the data: EURING2023 in Montpellier April 17-21","type":"blog"},{"authors":[],"categories":[],"content":"Quick \u0026amp; dirty text mining analysis of #ISEC2022 abstracts using topic modelling. Nothing out of the ordinary. Would be curious to assess time trends in topics w/ longitudinal analysis @chrissuthy 😇😉\nData and code: https://t.co/E2x7avw0E1\nPaper: https://t.co/l3ChAIa6Vi pic.twitter.com/NUF1FGm3Lv\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 4, 2022 ","date":1656892800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1656892800,"objectID":"13387cf60284fc9cdcff4ea39409f6cd","permalink":"https://oliviergimenez.github.io/blog/2022-07-04-quick-and-dirty-text-mining-analysis-of-isec2022-abstracts/","publishdate":"2022-07-04T00:00:00Z","relpermalink":"/blog/2022-07-04-quick-and-dirty-text-mining-analysis-of-isec2022-abstracts/","section":"blog","summary":"Quick \u0026amp; dirty text mining analysis of #ISEC2022 abstracts using topic modelling. Nothing out of the ordinary. Would be curious to assess time trends in topics w/ longitudinal analysis @chrissuthy 😇😉","tags":[],"title":"Quick and dirty text mining analysis of ISEC2022 abstracts","type":"blog"},{"authors":[],"categories":[],"content":"Would you like to get the most out of your capture-recapture data?\nWe run a workshop on multisite/state/event HMMs models in #Rstats Nimble w/ Daniel Turek @R_nimble at #ISEC2022 @ISEC_stats_ecol\nWebsite w/ slides \u0026amp; codes: https://t.co/gI1ccZO3O9\nVideo: https://t.co/zA6nh6Ttih pic.twitter.com/ejyE2tCw5o\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 1, 2022 ","date":1656633600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1656633600,"objectID":"0d30c499b67b1892f4dc7dcbd9a91251","permalink":"https://oliviergimenez.github.io/blog/2022-07-01-workshop-at-isec2022-on-capture-recapture-hmms-and-nimble/","publishdate":"2022-07-01T00:00:00Z","relpermalink":"/blog/2022-07-01-workshop-at-isec2022-on-capture-recapture-hmms-and-nimble/","section":"blog","summary":"Would you like to get the most out of your capture-recapture data?\nWe run a workshop on multisite/state/event HMMs models in #Rstats Nimble w/ Daniel Turek @R_nimble at #ISEC2022 @ISEC_stats_ecol","tags":["nimble","capture-recapture","hideen-markov-models","hmm"],"title":"Workshop at ISEC2022 on capture-recapture HMMs and Nimble","type":"blog"},{"authors":[],"categories":[],"content":"Thrilled to see our work on Mediterranean monk seal out in @AnimalConserv 🦭\nWe show a significant range recovery of the species in Greece 🇬🇷 using #CitizenScience \u0026amp; occupancy modelshttps://t.co/ILYdzArZq7\nCheck out NGO MOm amazing work https://t.co/hM8BhfRQ7w\nDM ➡️ PDF 😉 pic.twitter.com/ILqnFvMypL\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 30, 2022 ","date":1656547200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1656547200,"objectID":"74c4442d5b4312cf358991d71cb78136","permalink":"https://oliviergimenez.github.io/blog/2022-06-30-new-paper-on-monk-seal-recovery-in-greece/","publishdate":"2022-06-30T00:00:00Z","relpermalink":"/blog/2022-06-30-new-paper-on-monk-seal-recovery-in-greece/","section":"blog","summary":"Thrilled to see our work on Mediterranean monk seal out in @AnimalConserv 🦭\nWe show a significant range recovery of the species in Greece 🇬🇷 using #CitizenScience \u0026amp; occupancy modelshttps://t.co/ILYdzArZq7","tags":["dynamic-occupancy","monk-seals","citizen-science"],"title":"New paper on monk seal recovery in Greece","type":"blog"},{"authors":[],"categories":[],"content":"I\u0026#39;m always reluctant to give career advice, cause what worked for me won\u0026#39;t work for others, and there is obviously a survivorship bias. For what it\u0026#39;s worth, my 10 pieces of advice to early career researchers in statistical ecology #ISEC2022 🙏 @BayesRuth for the Q 😉\n🧵👇 https://t.co/JJEHFFOAc9\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 28, 2022 ","date":1656460800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1656460800,"objectID":"610f48360d290731338b120bc6ff1da0","permalink":"https://oliviergimenez.github.io/blog/2022-06-29-career-advice-to-early-career-researchers/","publishdate":"2022-06-29T00:00:00Z","relpermalink":"/blog/2022-06-29-career-advice-to-early-career-researchers/","section":"blog","summary":"I\u0026#39;m always reluctant to give career advice, cause what worked for me won\u0026#39;t work for others, and there is obviously a survivorship bias. For what it\u0026#39;s worth, my 10 pieces of advice to early career researchers in statistical ecology #ISEC2022 🙏 @BayesRuth for the Q 😉","tags":["ECR","academic-life"],"title":"Career advice to Early Career Researchers","type":"blog"},{"authors":[],"categories":[],"content":"Statistical ecology 🌍📊, hidden Markov models 🛰️ and large carnivores 🐺😺🐻 Soooo happy to share our research in this plenary talk at #ISEC2022 @ISEC_stats_ecol 🤩🍾\nslides and text: https://t.co/BOtZHWU56L\nvideo: https://t.co/6Qn7i05uae pic.twitter.com/vrSYeo9HYF\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 28, 2022 ","date":1656374400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1656374400,"objectID":"3d001a111f047e215e4d4f48e36c6d77","permalink":"https://oliviergimenez.github.io/blog/2022-06-28-my-plenary-talk-at-isec2022-on-statistical-ecology-hidden-markov-models-and-large-carnivores/","publishdate":"2022-06-28T00:00:00Z","relpermalink":"/blog/2022-06-28-my-plenary-talk-at-isec2022-on-statistical-ecology-hidden-markov-models-and-large-carnivores/","section":"blog","summary":"Statistical ecology 🌍📊, hidden Markov models 🛰️ and large carnivores 🐺😺🐻 Soooo happy to share our research in this plenary talk at #ISEC2022 @ISEC_stats_ecol 🤩🍾\nslides and text: https://t.co/BOtZHWU56L\nvideo: https://t.","tags":["hidden-markov-models","hmm","dynamic-occupancy","capture-recapture","conference"],"title":"My plenary talk at ISEC2022 on Statistical ecology, hidden Markov models and large carnivores.","type":"blog"},{"authors":[],"categories":[],"content":"Yesterday I created my 200th repo on @github 😇 It\u0026#39;s been quite an adventure to twist my old habits \u0026amp; switch to #ReproducibleResearch but it\u0026#39;s definitely worth it 🤯 Plenty of help out there like @living_norway 🤩 If useful, I also have a 1-day workshop https://t.co/7I6rU61nEg pic.twitter.com/cGDjaA6iTv\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 9, 2022 ","date":1654732800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1654732800,"objectID":"c52fe494c94078f74b0928f0afb551ed","permalink":"https://oliviergimenez.github.io/blog/2022-06-09-reproducible-research-and-github/","publishdate":"2022-06-09T00:00:00Z","relpermalink":"/blog/2022-06-09-reproducible-research-and-github/","section":"blog","summary":"Yesterday I created my 200th repo on @github 😇 It\u0026#39;s been quite an adventure to twist my old habits \u0026amp; switch to #ReproducibleResearch but it\u0026#39;s definitely worth it 🤯 Plenty of help out there like @living_norway 🤩 If useful, I also have a 1-day workshop https://t.","tags":["github","git","reproducible-research"],"title":"Reproducible research and GitHub","type":"blog"},{"authors":[],"categories":[],"content":"Check it out, this is huge 🤯 Awesome initiative led by @FredBarraquand Let\u0026#39;s get ourselves a 💎 #OpenAccess journal in #StatisticalEcology 🤩 Free for readers and for authors https://t.co/HmDhrCLsaN https://t.co/AI7Jjunoek pic.twitter.com/Aywk2ucusQ\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 8, 2022 ","date":1654646400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1654646400,"objectID":"f06e40d805610dbcbe32dcc7a0258141","permalink":"https://oliviergimenez.github.io/blog/2022-06-08-a-proposal-for-a-diamond-open-access-journal-in-statistical-ecology/","publishdate":"2022-06-08T00:00:00Z","relpermalink":"/blog/2022-06-08-a-proposal-for-a-diamond-open-access-journal-in-statistical-ecology/","section":"blog","summary":"Check it out, this is huge 🤯 Awesome initiative led by @FredBarraquand Let\u0026#39;s get ourselves a 💎 #OpenAccess journal in #StatisticalEcology 🤩 Free for readers and for authors https://t.co/HmDhrCLsaN https://t.co/AI7Jjunoek pic.","tags":["open-access","open-science","statistical-ecology"],"title":"A proposal for a diamond Open Access journal in Statistical Ecology","type":"blog"},{"authors":[],"categories":[],"content":"Because of my old age I guess, I\u0026#39;m often asked how I switched from base R #RStats to #tidyverse. I learnt by watching @drob screencasts 🎥 in which he analyses #TidyTuesday data he has never seen before (!). Here is a list of my tips https://t.co/VFAaDz8onE in case it\u0026#39;s useful. pic.twitter.com/4Emq2J2qQL\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 1, 2022 ","date":1654041600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1654041600,"objectID":"357e908f6c1ad3f64bd6631590cfc614","permalink":"https://oliviergimenez.github.io/blog/2022-06-01-tips-and-tricks-in-tidyverse/","publishdate":"2022-06-01T00:00:00Z","relpermalink":"/blog/2022-06-01-tips-and-tricks-in-tidyverse/","section":"blog","summary":"Because of my old age I guess, I\u0026#39;m often asked how I switched from base R #RStats to #tidyverse. I learnt by watching @drob screencasts 🎥 in which he analyses #TidyTuesday data he has never seen before (!","tags":["tidyverse","dplyr","ggplot"],"title":"Tips and tricks in Tidyverse","type":"blog"},{"authors":[],"categories":[],"content":"A Montpellier, @oaggimenez présente le fonctionnement du DIPEE Occitanie-Est qui rassemble plusieurs laboratoires @UmrAmap @cefemontpellier @criobe_pf @isemevol @umrMARBEC @MIVEGEC @ihpe_lab @EcotronCNRS Pour en savoir plus sur ce DIPEE 👇https://t.co/QvGniMkjg2 pic.twitter.com/7ekUx3vwu5\n\u0026mdash; L\u0026#39;écologie au CNRS (@INEE_CNRS) May 31, 2022 ","date":1653955200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1653955200,"objectID":"7b7075e6832a56c2276e329554b853fc","permalink":"https://oliviergimenez.github.io/blog/2022-05-31-semaine-ecologie-environnement-biodiversite-du-cnrs-a-montpellier/","publishdate":"2022-05-31T00:00:00Z","relpermalink":"/blog/2022-05-31-semaine-ecologie-environnement-biodiversite-du-cnrs-a-montpellier/","section":"blog","summary":"A Montpellier, @oaggimenez présente le fonctionnement du DIPEE Occitanie-Est qui rassemble plusieurs laboratoires @UmrAmap @cefemontpellier @criobe_pf @isemevol @umrMARBEC @MIVEGEC @ihpe_lab @EcotronCNRS Pour en savoir plus sur ce DIPEE 👇https://t.co/QvGniMkjg2 pic.twitter.com/7ekUx3vwu5","tags":["cnrs","conference"],"title":"Semaine écologie environnement biodiversité du CNRS à Montpellier","type":"blog"},{"authors":[],"categories":[],"content":"It\u0026#39;s real 😇🤩🥳 https://t.co/HdWol2H6PJ pic.twitter.com/NUSXtS1fQy\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) May 24, 2022 ","date":1653350400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1653350400,"objectID":"8a6d3af2659a8384cefa8de9b4a567ef","permalink":"https://oliviergimenez.github.io/blog/2022-05-24-our-new-book-on-statistical-approaches-for-hiddens-variables-in-ecology/","publishdate":"2022-05-24T00:00:00Z","relpermalink":"/blog/2022-05-24-our-new-book-on-statistical-approaches-for-hiddens-variables-in-ecology/","section":"blog","summary":"It\u0026#39;s real 😇🤩🥳 https://t.co/HdWol2H6PJ pic.twitter.com/NUSXtS1fQy\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) May 24, 2022 ","tags":["hmm","hidden-markov-models","latent-variables","state-space-models","statistical-ecology"],"title":"Our new book on statistical approaches for hiddens variables in ecology","type":"blog"},{"authors":[],"categories":[],"content":"🔧🚧 I have updated my quick introduction to data visualisation w/ R 📦 ggplot2 🥳\nIn case it\u0026#39;s useful 😉▶️ https://t.co/IWNesVjwqq ◀️\nw/ practicals https://t.co/YYkykGeWf1 \u0026amp; solutions https://t.co/NCQYsw8QCg Cartoons @allison_horst 💜#ggplot #tidyverse #dataviz #RStats pic.twitter.com/RlXFT9F0Tw\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) May 23, 2022 ","date":1653264000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1653264000,"objectID":"e1e6487655975f75170253a1eb37d2ea","permalink":"https://oliviergimenez.github.io/blog/2022-05-23-my-quick-introduction-to-data-visualisation-with-r/","publishdate":"2022-05-23T00:00:00Z","relpermalink":"/blog/2022-05-23-my-quick-introduction-to-data-visualisation-with-r/","section":"blog","summary":"🔧🚧 I have updated my quick introduction to data visualisation w/ R 📦 ggplot2 🥳\nIn case it\u0026#39;s useful 😉▶️ https://t.co/IWNesVjwqq ◀️\nw/ practicals https://t.co/YYkykGeWf1 \u0026amp; solutions https://t.co/NCQYsw8QCg Cartoons @allison_horst 💜#ggplot #tidyverse #dataviz #RStats pic.","tags":["ggplot","dataviz","data-visualisation"],"title":"My quick introduction to data visualisation with R","type":"blog"},{"authors":[],"categories":[],"content":"🔧🚧 Intro on data wrangling w/ R 📦 dplyr - import data, pipe, pivot, join, filter, count, work w/ columns, dates, char, factors, etc #tidyverse\n▶️ https://t.co/Vs2TbYhbBj ◀️\nw/ practicals https://t.co/9ktKPp9wfW \u0026amp; solutions https://t.co/YPwibL2w31 Cartoons @allison_horst 💜 pic.twitter.com/ZMSfxYFKWq\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) May 17, 2022 ","date":1652745600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1652745600,"objectID":"904a99402fa0b6816cd97671aee28761","permalink":"https://oliviergimenez.github.io/blog/2022-05-17-introduction-on-data-wrangling-with-r/","publishdate":"2022-05-17T00:00:00Z","relpermalink":"/blog/2022-05-17-introduction-on-data-wrangling-with-r/","section":"blog","summary":"🔧🚧 Intro on data wrangling w/ R 📦 dplyr - import data, pipe, pivot, join, filter, count, work w/ columns, dates, char, factors, etc #tidyverse\n▶️ https://t.co/Vs2TbYhbBj ◀️\nw/ practicals https://t.","tags":["dplyr","R","data-wrangling"],"title":"Introduction on data wrangling with R","type":"blog"},{"authors":[],"categories":[],"content":"😇 I updated the slides of my introduction to #GIS and #mapping in #rstats using the #sf package and 🐻 🗺️ in the #pyrenees 🏔️ 🧑‍🏫 Slides: https://t.co/mO4Dg8l1H5\n🧑‍💻 Code: https://t.co/X8fDKcrbFR#rspatial #spatial #ggplot2 #tidyverse #dataviz pic.twitter.com/mSsL6Qt7IS\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) May 12, 2022 ","date":1652313600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1652313600,"objectID":"b1711fff361df7eef2af6caa4f9a0663","permalink":"https://oliviergimenez.github.io/blog/2022-05-12-introduction-to-gis-and-mapping-in-r/","publishdate":"2022-05-12T00:00:00Z","relpermalink":"/blog/2022-05-12-introduction-to-gis-and-mapping-in-r/","section":"blog","summary":"😇 I updated the slides of my introduction to #GIS and #mapping in #rstats using the #sf package and 🐻 🗺️ in the #pyrenees 🏔️ 🧑‍🏫 Slides: https://t.co/mO4Dg8l1H5\n🧑‍💻 Code: https://t.","tags":["sf","R","gis","sig"],"title":"Introduction to GIS and mapping in R","type":"blog"},{"authors":[],"categories":[],"content":"Happy to share lecture slides \u0026amp; R codes of an earlier version of the workshop 👇 we ran back in 2014 🤯 w JD Lebreton \u0026amp; @KoonsLab ➡️ https://t.co/XdcCQ560Pz #RStats For video 📽️📺, check out our workshop on pop dynamics https://t.co/jlTE4iF5zk w @SarahCubaynes \u0026amp; @abesnardEPHE https://t.co/JFgS1GOU0V\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) May 4, 2022 ","date":1651622400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1651622400,"objectID":"47c5f57a24998da50317c378568d6dc0","permalink":"https://oliviergimenez.github.io/blog/2022-05-04-matrix-population-models-lecture-slides-and-r-codes-for-a-workshop/","publishdate":"2022-05-04T00:00:00Z","relpermalink":"/blog/2022-05-04-matrix-population-models-lecture-slides-and-r-codes-for-a-workshop/","section":"blog","summary":"Happy to share lecture slides \u0026amp; R codes of an earlier version of the workshop 👇 we ran back in 2014 🤯 w JD Lebreton \u0026amp; @KoonsLab ➡️ https://t.co/XdcCQ560Pz #RStats For video 📽️📺, check out our workshop on pop dynamics https://t.","tags":["matrix-population-models","caswell","R","open-science"],"title":"Matrix population models: Lecture slides and R codes for a workshop","type":"blog"},{"authors":[],"categories":[],"content":"Useful papers on GLMMs 🤩🥳@bolkerb https://t.co/HX41N05rwK@BobOHara https://t.co/c8fVvxG7ru@mattjsilk https://t.co/Lq3xAf5rMh@XavHarrison https://t.co/Nx2w81N34F@LisaDeBruine @dalejbarr https://t.co/QOrQorDD89\nSee also GLMM FAQ by @bolkerb \u0026amp; others https://t.co/s9AmINlleU pic.twitter.com/oGYC3GMw99\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) May 3, 2022 ","date":1651536000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1651536000,"objectID":"d91869d0e2c8305969c609e41c4595d4","permalink":"https://oliviergimenez.github.io/blog/2022-05-03-useful-papers-on-glmms/","publishdate":"2022-05-03T00:00:00Z","relpermalink":"/blog/2022-05-03-useful-papers-on-glmms/","section":"blog","summary":"Useful papers on GLMMs 🤩🥳@bolkerb https://t.co/HX41N05rwK@BobOHara https://t.co/c8fVvxG7ru@mattjsilk https://t.co/Lq3xAf5rMh@XavHarrison https://t.co/Nx2w81N34F@LisaDeBruine @dalejbarr https://t.co/QOrQorDD89\nSee also GLMM FAQ by @bolkerb \u0026amp; others https://t.co/s9AmINlleU pic.twitter.com/oGYC3GMw99\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) May 3, 2022 ","tags":["glmm"],"title":"Useful papers on GLMMs","type":"blog"},{"authors":null,"categories":null,"content":"Here I illustrate how to fit GLMMs with the R package brms, and compare to Jags and lme4.\nMotivation I regularly give a course on Bayesian statistics with R for non-specialists. To illustrate the course, we analyse data with generalized linear, often mixed, models or GLMMs.\nSo far, I\u0026rsquo;ve been using Jags to fit these models. This requires some programming skills, like e.g. coding a loop, to be able to write down the model likelihood. Although students learn a lot from going through that process, it can be daunting.\nThis year, I thought I\u0026rsquo;d show them the R package brms developed by Paul-Christian Bürkner. In brief, brms allows fitting GLMMs (but not only) in a lme4-like syntax within the Bayesian framework and MCMC methods with Stan.\nI\u0026rsquo;m not a Stan user, but it doesn\u0026rsquo;t matter. The vignettes were more than enough to get me started. I also recommend the list of blog posts about brms.\nFirst things first, we load the packages we will need:\nlibrary(brms) library(posterior) # tools for working with posterior and prior distributions library(R2jags) # run Jags from within R library(lme4) # fit GLMM in frequentist framework Beta-Binomial model The first example is about a survival experiment. We captured, marked and released 57 individuals (total) out of which 19 made it through the year (alive).\ndat \u0026lt;- data.frame(alive = 19, total = 57) dat An obvious estimate of the probability of a binomial is the proportion of cases, $19/57 = 0.3333333$ here, but let\u0026rsquo;s use R built-in functions for the sake of illustration.\nMaximum-likelihood estimation To get an estimate of survival, we fit a logistic regression using the glm() function. The data are grouped (or aggregated), so we need to specify the number of alive and dead individuals as a two-column matrix on the left hand side of the model formula:\nmle \u0026lt;- glm(cbind(alive, total - alive) ~ 1, family = binomial(\u0026quot;logit\u0026quot;), # family = binomial(\u0026quot;identity\u0026quot;) would be more straightforward data = dat) On the logit scale, survival is estimated at:\ncoef(mle) # logit scale ## (Intercept) ## -0.6931472 After back-transformation using the reciprocal logit, we obtain:\nplogis(coef(mle)) ## (Intercept) ## 0.3333333 So far, so good.\nBayesian analysis with Jags In Jags, you can fit this model with:\n# model betabin \u0026lt;- function(){ alive ~ dbin(survival, total) # binomial likelihood logit(survival) \u0026lt;- beta # logit link beta ~ dnorm(0, 1/1.5) # prior } # data datax \u0026lt;- list(total = 57, alive = 19) # initial values inits \u0026lt;- function() list(beta = rnorm(1, 0, sd = 1.5)) # parameter to monitor params \u0026lt;- c(\u0026quot;survival\u0026quot;) # run jags bayes.jags \u0026lt;- jags(data = datax, inits = inits, parameters.to.save = params, model.file = betabin, n.chains = 2, n.iter = 5000, n.burnin = 1000, n.thin = 1) ## Compiling model graph ## Resolving undeclared variables ## Allocating nodes ## Graph information: ## Observed stochastic nodes: 1 ## Unobserved stochastic nodes: 1 ## Total graph size: 8 ## ## Initializing model # display results bayes.jags ## Inference for Bugs model at \u0026quot;/var/folders/r7/j0wqj1k95vz8w44sdxzm986c0000gn/T//RtmpP7MZ1A/modela81d72dd23e2.txt\u0026quot;, fit using jags, ## 2 chains, each with 5000 iterations (first 1000 discarded) ## n.sims = 8000 iterations saved ## mu.vect sd.vect 2.5% 25% 50% 75% 97.5% Rhat n.eff ## survival 0.342 0.061 0.228 0.300 0.341 0.382 0.466 1.001 2600 ## deviance 5.354 1.390 4.388 4.486 4.819 5.662 9.333 1.001 8000 ## ## For each parameter, n.eff is a crude measure of effective sample size, ## and Rhat is the potential scale reduction factor (at convergence, Rhat=1). ## ## DIC info (using the rule, pD = var(deviance)/2) ## pD = 1.0 and DIC = 6.3 ## DIC is an estimate of expected predictive error (lower deviance is better). Bayesian analysis with brms In brms, you write:\nbayes.brms \u0026lt;- brm(alive | trials(total) ~ 1, family = binomial(\u0026quot;logit\u0026quot;), # binomial(\u0026quot;identity\u0026quot;) would be more straightforward data = dat, chains = 2, # nb of chains iter = 5000, # nb of iterations, including burnin warmup = 1000, # burnin thin = 1) # thinning ## Running /Library/Frameworks/R.framework/Resources/bin/R CMD SHLIB foo.c ## clang -mmacosx-version-min=10.13 -I\u0026quot;/Library/Frameworks/R.framework/Resources/include\u0026quot; -DNDEBUG -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/Rcpp/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/unsupported\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/BH/include\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/src/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppParallel/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/rstan/include\u0026quot; -DEIGEN_NO_DEBUG -DBOOST_DISABLE_ASSERTS -DBOOST_PENDING_INTEGER_LOG2_HPP -DSTAN_THREADS -DBOOST_NO_AUTO_PTR -include '/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp' -D_REENTRANT -DRCPP_PARALLEL_USE_TBB=1 -I/usr/local/include -fPIC -Wall -g -O2 -c foo.c -o foo.o ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:88: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:1: error: unknown type name 'namespace' ## namespace Eigen { ## ^ ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:16: error: expected ';' after top level declarator ## namespace Eigen { ## ^ ## ; ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:96:10: fatal error: 'complex' file not found ## #include \u0026lt;complex\u0026gt; ## ^~~~~~~~~ ## 3 errors generated. ## make: *** [foo.o] Error 1 ## ## SAMPLING FOR MODEL '782790eabbd10296f779d85fa03cf1e5' NOW (CHAIN 1). ## Chain 1: ## Chain 1: Gradient evaluation took 2e-05 seconds ## Chain 1: 1000 transitions using 10 leapfrog steps per transition would take 0.2 seconds. ## Chain 1: Adjust your expectations accordingly! ## Chain 1: ## Chain 1: ## Chain 1: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 1: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 1: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 1: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 1: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 1: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 1: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 1: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 1: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 1: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 1: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 1: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 1: ## Chain 1: Elapsed Time: 0.010358 seconds (Warm-up) ## Chain 1: 0.046464 seconds (Sampling) ## Chain 1: 0.056822 seconds (Total) ## Chain 1: ## ## SAMPLING FOR MODEL '782790eabbd10296f779d85fa03cf1e5' NOW (CHAIN 2). ## Chain 2: ## Chain 2: Gradient evaluation took 1.3e-05 seconds ## Chain 2: 1000 transitions using 10 leapfrog steps per transition would take 0.13 seconds. ## Chain 2: Adjust your expectations accordingly! ## Chain 2: ## Chain 2: ## Chain 2: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 2: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 2: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 2: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 2: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 2: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 2: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 2: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 2: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 2: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 2: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 2: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 2: ## Chain 2: Elapsed Time: 0.009763 seconds (Warm-up) ## Chain 2: 0.035949 seconds (Sampling) ## Chain 2: 0.045712 seconds (Total) ## Chain 2: You can display the results:\nbayes.brms ## Family: binomial ## Links: mu = logit ## Formula: alive | trials(total) ~ 1 ## Data: dat (Number of observations: 1) ## Draws: 2 chains, each with iter = 5000; warmup = 1000; thin = 1; ## total post-warmup draws = 8000 ## ## Population-Level Effects: ## Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS ## Intercept -0.69 0.28 -1.27 -0.14 1.00 3213 3572 ## ## Draws were sampled using sampling(NUTS). For each parameter, Bulk_ESS ## and Tail_ESS are effective sample size measures, and Rhat is the potential ## scale reduction factor on split chains (at convergence, Rhat = 1). And visualize the posterior density and trace of survival (on the logit scale):\nplot(bayes.brms) To get survival on the $[0,1]$ scale, we extract the MCMC values, then apply the reciprocal logit function to each of these values, and summarize its posterior distribution:\ndraws_fit \u0026lt;- as_draws_matrix(bayes.brms) draws_fit ## # A draws_matrix: 4000 iterations, 2 chains, and 2 variables ## variable ## draw b_Intercept lp__ ## 1 -0.70 -4.2 ## 2 -0.67 -4.2 ## 3 -0.86 -4.4 ## 4 -0.80 -4.3 ## 5 -1.32 -6.6 ## 6 -0.57 -4.2 ## 7 -0.51 -4.4 ## 8 -0.56 -4.3 ## 9 -0.69 -4.2 ## 10 -0.41 -4.6 ## # ... with 7990 more draws summarize_draws(plogis(draws_fit[,1])) What is the prior used by default?\nprior_summary(bayes.brms) What if I want to use the same prior as in Jags instead?\nnlprior \u0026lt;- prior(normal(0, 1.5), class = \u0026quot;Intercept\u0026quot;) # new prior bayes.brms \u0026lt;- brm(alive | trials(total) ~ 1, family = binomial(\u0026quot;logit\u0026quot;), data = dat, prior = nlprior, # set prior by hand chains = 2, # nb of chains iter = 5000, # nb of iterations, including burnin warmup = 1000, # burnin thin = 1) ## Running /Library/Frameworks/R.framework/Resources/bin/R CMD SHLIB foo.c ## clang -mmacosx-version-min=10.13 -I\u0026quot;/Library/Frameworks/R.framework/Resources/include\u0026quot; -DNDEBUG -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/Rcpp/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/unsupported\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/BH/include\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/src/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppParallel/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/rstan/include\u0026quot; -DEIGEN_NO_DEBUG -DBOOST_DISABLE_ASSERTS -DBOOST_PENDING_INTEGER_LOG2_HPP -DSTAN_THREADS -DBOOST_NO_AUTO_PTR -include '/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp' -D_REENTRANT -DRCPP_PARALLEL_USE_TBB=1 -I/usr/local/include -fPIC -Wall -g -O2 -c foo.c -o foo.o ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:88: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:1: error: unknown type name 'namespace' ## namespace Eigen { ## ^ ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:16: error: expected ';' after top level declarator ## namespace Eigen { ## ^ ## ; ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:96:10: fatal error: 'complex' file not found ## #include \u0026lt;complex\u0026gt; ## ^~~~~~~~~ ## 3 errors generated. ## make: *** [foo.o] Error 1 ## ## SAMPLING FOR MODEL '59dcc499a2b4c8a1b2c8f4e80e70fcce' NOW (CHAIN 1). ## Chain 1: ## Chain 1: Gradient evaluation took 1.8e-05 seconds ## Chain 1: 1000 transitions using 10 leapfrog steps per transition would take 0.18 seconds. ## Chain 1: Adjust your expectations accordingly! ## Chain 1: ## Chain 1: ## Chain 1: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 1: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 1: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 1: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 1: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 1: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 1: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 1: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 1: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 1: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 1: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 1: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 1: ## Chain 1: Elapsed Time: 0.00875 seconds (Warm-up) ## Chain 1: 0.03345 seconds (Sampling) ## Chain 1: 0.0422 seconds (Total) ## Chain 1: ## ## SAMPLING FOR MODEL '59dcc499a2b4c8a1b2c8f4e80e70fcce' NOW (CHAIN 2). ## Chain 2: ## Chain 2: Gradient evaluation took 5e-06 seconds ## Chain 2: 1000 transitions using 10 leapfrog steps per transition would take 0.05 seconds. ## Chain 2: Adjust your expectations accordingly! ## Chain 2: ## Chain 2: ## Chain 2: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 2: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 2: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 2: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 2: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 2: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 2: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 2: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 2: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 2: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 2: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 2: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 2: ## Chain 2: Elapsed Time: 0.009788 seconds (Warm-up) ## Chain 2: 0.036797 seconds (Sampling) ## Chain 2: 0.046585 seconds (Total) ## Chain 2: Double-check the prior that was used:\nprior_summary(bayes.brms) Logistic regression with covariates In this example, we ask whether annual variation in white stork breeding success can be explained by rainfall in the wintering area. Breeding success is measured by the ratio of the number of chicks over the number of pairs.\nnbchicks \u0026lt;- c(151,105,73,107,113,87,77,108,118,122,112,120,122,89,69,71,53,41,53,31,35,14,18) nbpairs \u0026lt;- c(173,164,103,113,122,112,98,121,132,136,133,137,145,117,90,80,67,54,58,39,42,23,23) rain \u0026lt;- c(67,52,88,61,32,36,72,43,92,32,86,28,57,55,66,26,28,96,48,90,86,78,87) dat \u0026lt;- data.frame(nbchicks = nbchicks, nbpairs = nbpairs, rain = (rain - mean(rain))/sd(rain)) # standardized rainfall The data are grouped.\nMaximum-likelihood estimation You can get maximum likelihood estimates with:\nmle \u0026lt;- glm(cbind(nbchicks, nbpairs - nbchicks) ~ rain, family = binomial(\u0026quot;logit\u0026quot;), data = dat) summary(mle) ## ## Call: ## glm(formula = cbind(nbchicks, nbpairs - nbchicks) ~ rain, family = binomial(\u0026quot;logit\u0026quot;), ## data = dat) ## ## Deviance Residuals: ## Min 1Q Median 3Q Max ## -5.9630 -1.3354 0.3929 1.6168 3.8800 ## ## Coefficients: ## Estimate Std. Error z value Pr(\u0026gt;|z|) ## (Intercept) 1.55436 0.05565 27.931 \u0026lt;2e-16 *** ## rain -0.14857 0.05993 -2.479 0.0132 * ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 ## ## (Dispersion parameter for binomial family taken to be 1) ## ## Null deviance: 109.25 on 22 degrees of freedom ## Residual deviance: 103.11 on 21 degrees of freedom ## AIC: 205.84 ## ## Number of Fisher Scoring iterations: 4 There is a negative effect of rainfall on breeding success:\nvisreg::visreg(mle, scale = \u0026quot;response\u0026quot;) Bayesian analysis with Jags In Jags, we can fit the same model:\n# data dat \u0026lt;- list(N = 23, # nb of years nbchicks = nbchicks, nbpairs = nbpairs, rain = (rain - mean(rain))/sd(rain)) # standardized rainfall # define model logistic \u0026lt;- function() { # likelihood for(i in 1:N){ # loop over years nbchicks[i] ~ dbin(p[i], nbpairs[i]) # binomial likelihood logit(p[i]) \u0026lt;- a + b.rain * rain[i] # prob success is linear fn of rainfall on logit scale } # priors a ~ dnorm(0,0.01) # intercept b.rain ~ dnorm(0,0.01) # slope for rainfall } # function that generates initial values inits \u0026lt;- function() list(a = rnorm(n = 1, mean = 0, sd = 5), b.rain = rnorm(n = 1, mean = 0, sd = 5)) # specify parameters that need to be estimated parameters \u0026lt;- c(\u0026quot;a\u0026quot;, \u0026quot;b.rain\u0026quot;) # run Jags bayes.jags \u0026lt;- jags(data = dat, inits = inits, parameters.to.save = parameters, model.file = logistic, n.chains = 2, n.iter = 5000, # includes burn-in! n.burnin = 1000) ## Compiling model graph ## Resolving undeclared variables ## Allocating nodes ## Graph information: ## Observed stochastic nodes: 23 ## Unobserved stochastic nodes: 2 ## Total graph size: 134 ## ## Initializing model # display results bayes.jags ## Inference for Bugs model at \u0026quot;/var/folders/r7/j0wqj1k95vz8w44sdxzm986c0000gn/T//RtmpP7MZ1A/modela81d1b01bd3f.txt\u0026quot;, fit using jags, ## 2 chains, each with 5000 iterations (first 1000 discarded), n.thin = 4 ## n.sims = 2000 iterations saved ## mu.vect sd.vect 2.5% 25% 50% 75% 97.5% Rhat n.eff ## a 1.546 0.163 1.439 1.514 1.552 1.588 1.657 1.182 2000 ## b.rain -0.152 0.121 -0.275 -0.190 -0.148 -0.106 -0.033 1.129 2000 ## deviance 213.612 330.910 201.899 202.434 203.197 204.689 209.169 1.121 2000 ## ## For each parameter, n.eff is a crude measure of effective sample size, ## and Rhat is the potential scale reduction factor (at convergence, Rhat=1). ## ## DIC info (using the rule, pD = var(deviance)/2) ## pD = 54753.9 and DIC = 54967.5 ## DIC is an estimate of expected predictive error (lower deviance is better). Bayesian analysis with brms With brms, we write:\nbayes.brms \u0026lt;- brm(nbchicks | trials(nbpairs) ~ rain, family = binomial(\u0026quot;logit\u0026quot;), data = dat, chains = 2, # nb of chains iter = 5000, # nb of iterations, including burnin warmup = 1000, # burnin thin = 1) ## Running /Library/Frameworks/R.framework/Resources/bin/R CMD SHLIB foo.c ## clang -mmacosx-version-min=10.13 -I\u0026quot;/Library/Frameworks/R.framework/Resources/include\u0026quot; -DNDEBUG -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/Rcpp/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/unsupported\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/BH/include\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/src/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppParallel/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/rstan/include\u0026quot; -DEIGEN_NO_DEBUG -DBOOST_DISABLE_ASSERTS -DBOOST_PENDING_INTEGER_LOG2_HPP -DSTAN_THREADS -DBOOST_NO_AUTO_PTR -include '/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp' -D_REENTRANT -DRCPP_PARALLEL_USE_TBB=1 -I/usr/local/include -fPIC -Wall -g -O2 -c foo.c -o foo.o ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:88: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:1: error: unknown type name 'namespace' ## namespace Eigen { ## ^ ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:16: error: expected ';' after top level declarator ## namespace Eigen { ## ^ ## ; ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:96:10: fatal error: 'complex' file not found ## #include \u0026lt;complex\u0026gt; ## ^~~~~~~~~ ## 3 errors generated. ## make: *** [foo.o] Error 1 ## ## SAMPLING FOR MODEL '70a68f0c53365768853c2791038e6352' NOW (CHAIN 1). ## Chain 1: ## Chain 1: Gradient evaluation took 4.7e-05 seconds ## Chain 1: 1000 transitions using 10 leapfrog steps per transition would take 0.47 seconds. ## Chain 1: Adjust your expectations accordingly! ## Chain 1: ## Chain 1: ## Chain 1: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 1: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 1: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 1: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 1: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 1: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 1: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 1: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 1: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 1: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 1: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 1: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 1: ## Chain 1: Elapsed Time: 0.035213 seconds (Warm-up) ## Chain 1: 0.131663 seconds (Sampling) ## Chain 1: 0.166876 seconds (Total) ## Chain 1: ## ## SAMPLING FOR MODEL '70a68f0c53365768853c2791038e6352' NOW (CHAIN 2). ## Chain 2: ## Chain 2: Gradient evaluation took 1.4e-05 seconds ## Chain 2: 1000 transitions using 10 leapfrog steps per transition would take 0.14 seconds. ## Chain 2: Adjust your expectations accordingly! ## Chain 2: ## Chain 2: ## Chain 2: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 2: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 2: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 2: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 2: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 2: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 2: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 2: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 2: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 2: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 2: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 2: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 2: ## Chain 2: Elapsed Time: 0.033715 seconds (Warm-up) ## Chain 2: 0.15264 seconds (Sampling) ## Chain 2: 0.186355 seconds (Total) ## Chain 2: Display results:\nbayes.brms ## Family: binomial ## Links: mu = logit ## Formula: nbchicks | trials(nbpairs) ~ rain ## Data: dat (Number of observations: 23) ## Draws: 2 chains, each with iter = 5000; warmup = 1000; thin = 1; ## total post-warmup draws = 8000 ## ## Population-Level Effects: ## Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS ## Intercept 1.56 0.06 1.45 1.67 1.00 6907 5043 ## rain -0.15 0.06 -0.27 -0.03 1.00 7193 5067 ## ## Draws were sampled using sampling(NUTS). For each parameter, Bulk_ESS ## and Tail_ESS are effective sample size measures, and Rhat is the potential ## scale reduction factor on split chains (at convergence, Rhat = 1). Visualize:\nplot(bayes.brms) We can also calculate the posterior probability of the rainfall effect being below zero with the hypothesis() function:\nhypothesis(bayes.brms, 'rain \u0026lt; 0') ## Hypothesis Tests for class b: ## Hypothesis Estimate Est.Error CI.Lower CI.Upper Evid.Ratio Post.Prob Star ## 1 (rain) \u0026lt; 0 -0.15 0.06 -0.25 -0.05 116.65 0.99 * ## --- ## 'CI': 90%-CI for one-sided and 95%-CI for two-sided hypotheses. ## '*': For one-sided hypotheses, the posterior probability exceeds 95%; ## for two-sided hypotheses, the value tested against lies outside the 95%-CI. ## Posterior probabilities of point hypotheses assume equal prior probabilities. These results confirm the important negative effect of rainfall on breeding success.\nLinear mixed model In this example, we have several measurements for 33 Mediterranean plant species, specifically number of seeds and biomass. We ask whether there is a linear relationship between these two variables that would hold for all species.\nRead in data, directly from the course website:\ndf \u0026lt;- read_csv2(\u0026quot;https://raw.githubusercontent.com/oliviergimenez/bayesian-stats-with-R/master/slides/dat/VMG.csv\u0026quot;) df Use relevant format for columns species Sp and biomass Vm:\ndf$Sp \u0026lt;- as_factor(df$Sp) df$Vm \u0026lt;- as.numeric(df$Vm) Define numeric vector of species, for nested indexing in Jags:\nspecies \u0026lt;- as.numeric(df$Sp) Define response variable, number of seeds:\ny \u0026lt;- log(df$NGrTotest) Standardize explanatory variable, biomass\nx \u0026lt;- (df$Vm - mean(df$Vm))/sd(df$Vm) Now build dataset:\ndat \u0026lt;- data.frame(y = y, x = x, species = species) Maximum-likelihood estimation You can get maximum likelihood estimates for a linear regression of number of seeds on biomass with a species random effect on the intercept (partial pooling):\nmle \u0026lt;- lmer(y ~ x + (1 | species), data = dat) summary(mle) ## Linear mixed model fit by REML ['lmerMod'] ## Formula: y ~ x + (1 | species) ## Data: dat ## ## REML criterion at convergence: 2642.5 ## ## Scaled residuals: ## Min 1Q Median 3Q Max ## -4.6761 -0.2704 -0.0120 0.2900 7.3163 ## ## Random effects: ## Groups Name Variance Std.Dev. ## species (Intercept) 113.146 10.637 ## Residual 9.373 3.062 ## Number of obs: 488, groups: species, 33 ## ## Fixed effects: ## Estimate Std. Error t value ## (Intercept) 14.5258 1.8574 7.821 ## x 0.4779 0.2421 1.974 ## ## Correlation of Fixed Effects: ## (Intr) ## x 0.002 Bayesian analysis with Jags In Jags, you would use:\n# partial pooling model partial_pooling \u0026lt;- function(){ # likelihood for(i in 1:n){ y[i] ~ dnorm(mu[i], tau.y) mu[i] \u0026lt;- a[species[i]] + b * x[i] } for (j in 1:nbspecies){ a[j] ~ dnorm(mu.a, tau.a) } # priors tau.y \u0026lt;- 1 / (sigma.y * sigma.y) sigma.y ~ dunif(0,100) mu.a ~ dnorm(0,0.01) tau.a \u0026lt;- 1 / (sigma.a * sigma.a) sigma.a ~ dunif(0,100) b ~ dnorm(0,0.01) } # data mydata \u0026lt;- list(y = y, x = x, n = length(y), species = species, nbspecies = length(levels(df$Sp))) # initial values inits \u0026lt;- function() list(mu.a = rnorm(1,0,5), sigma.a = runif(0,0,10), b = rnorm(1,0,5), sigma.y = runif(0,0,10)) # parameters to monitor params \u0026lt;- c(\u0026quot;mu.a\u0026quot;, \u0026quot;sigma.a\u0026quot;, \u0026quot;b\u0026quot;, \u0026quot;sigma.y\u0026quot;) # call jags to fit model bayes.jags \u0026lt;- jags(data = mydata, inits = inits, parameters.to.save = params, model.file = partial_pooling, n.chains = 2, n.iter = 5000, n.burnin = 1000, n.thin = 1) ## Compiling model graph ## Resolving undeclared variables ## Allocating nodes ## Graph information: ## Observed stochastic nodes: 488 ## Unobserved stochastic nodes: 37 ## Total graph size: 2484 ## ## Initializing model bayes.jags ## Inference for Bugs model at \u0026quot;/var/folders/r7/j0wqj1k95vz8w44sdxzm986c0000gn/T//RtmpP7MZ1A/modela81d7639dc1c.txt\u0026quot;, fit using jags, ## 2 chains, each with 5000 iterations (first 1000 discarded) ## n.sims = 8000 iterations saved ## mu.vect sd.vect 2.5% 25% 50% 75% 97.5% Rhat ## b 0.476 0.244 -0.005 0.314 0.472 0.638 0.964 1.001 ## mu.a 14.025 1.916 10.146 12.774 14.009 15.299 17.760 1.001 ## sigma.a 11.094 1.465 8.696 10.039 10.934 11.952 14.330 1.001 ## sigma.y 3.072 0.101 2.884 3.002 3.069 3.138 3.277 1.001 ## deviance 2478.206 8.704 2463.096 2472.022 2477.521 2483.650 2497.107 1.001 ## n.eff ## b 8000 ## mu.a 8000 ## sigma.a 8000 ## sigma.y 8000 ## deviance 8000 ## ## For each parameter, n.eff is a crude measure of effective sample size, ## and Rhat is the potential scale reduction factor (at convergence, Rhat=1). ## ## DIC info (using the rule, pD = var(deviance)/2) ## pD = 37.9 and DIC = 2516.1 ## DIC is an estimate of expected predictive error (lower deviance is better). Bayesian analysis with brms In brms, you write:\nbayes.brms \u0026lt;- brm(y ~ x + (1 | species), data = dat, chains = 2, # nb of chains iter = 5000, # nb of iterations, including burnin warmup = 1000, # burnin thin = 1) ## Running /Library/Frameworks/R.framework/Resources/bin/R CMD SHLIB foo.c ## clang -mmacosx-version-min=10.13 -I\u0026quot;/Library/Frameworks/R.framework/Resources/include\u0026quot; -DNDEBUG -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/Rcpp/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/unsupported\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/BH/include\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/src/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppParallel/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/rstan/include\u0026quot; -DEIGEN_NO_DEBUG -DBOOST_DISABLE_ASSERTS -DBOOST_PENDING_INTEGER_LOG2_HPP -DSTAN_THREADS -DBOOST_NO_AUTO_PTR -include '/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp' -D_REENTRANT -DRCPP_PARALLEL_USE_TBB=1 -I/usr/local/include -fPIC -Wall -g -O2 -c foo.c -o foo.o ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:88: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:1: error: unknown type name 'namespace' ## namespace Eigen { ## ^ ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:16: error: expected ';' after top level declarator ## namespace Eigen { ## ^ ## ; ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:96:10: fatal error: 'complex' file not found ## #include \u0026lt;complex\u0026gt; ## ^~~~~~~~~ ## 3 errors generated. ## make: *** [foo.o] Error 1 ## ## SAMPLING FOR MODEL 'c7dcb37f6d55803a5216866e78a4a476' NOW (CHAIN 1). ## Chain 1: ## Chain 1: Gradient evaluation took 0.000102 seconds ## Chain 1: 1000 transitions using 10 leapfrog steps per transition would take 1.02 seconds. ## Chain 1: Adjust your expectations accordingly! ## Chain 1: ## Chain 1: ## Chain 1: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 1: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 1: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 1: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 1: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 1: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 1: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 1: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 1: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 1: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 1: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 1: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 1: ## Chain 1: Elapsed Time: 0.656886 seconds (Warm-up) ## Chain 1: 3.37389 seconds (Sampling) ## Chain 1: 4.03077 seconds (Total) ## Chain 1: ## ## SAMPLING FOR MODEL 'c7dcb37f6d55803a5216866e78a4a476' NOW (CHAIN 2). ## Chain 2: ## Chain 2: Gradient evaluation took 2.7e-05 seconds ## Chain 2: 1000 transitions using 10 leapfrog steps per transition would take 0.27 seconds. ## Chain 2: Adjust your expectations accordingly! ## Chain 2: ## Chain 2: ## Chain 2: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 2: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 2: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 2: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 2: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 2: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 2: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 2: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 2: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 2: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 2: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 2: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 2: ## Chain 2: Elapsed Time: 0.593596 seconds (Warm-up) ## Chain 2: 2.34578 seconds (Sampling) ## Chain 2: 2.93937 seconds (Total) ## Chain 2: Display the results:\nbayes.brms ## Family: gaussian ## Links: mu = identity; sigma = identity ## Formula: y ~ x + (1 | species) ## Data: dat (Number of observations: 488) ## Draws: 2 chains, each with iter = 5000; warmup = 1000; thin = 1; ## total post-warmup draws = 8000 ## ## Group-Level Effects: ## ~species (Number of levels: 33) ## Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS ## sd(Intercept) 10.74 1.39 8.45 13.78 1.00 427 943 ## ## Population-Level Effects: ## Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS ## Intercept 14.13 1.90 10.31 17.86 1.01 331 580 ## x 0.47 0.25 -0.00 0.97 1.00 2317 3382 ## ## Family Specific Parameters: ## Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS ## sigma 3.07 0.10 2.88 3.28 1.00 2667 3408 ## ## Draws were sampled using sampling(NUTS). For each parameter, Bulk_ESS ## and Tail_ESS are effective sample size measures, and Rhat is the potential ## scale reduction factor on split chains (at convergence, Rhat = 1). You can extract a block of fixed effects:\nsummary(bayes.brms)$fixed And a block of random effects:\nsummary(bayes.brms)$random ## $species ## Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS ## sd(Intercept) 10.74273 1.388061 8.454543 13.776 1.001916 426.9568 943.1178 And visualize:\nplot(bayes.brms) GLMM with Poisson distribution This example is from Jason Matthiopoulos\u0026rsquo; excellent book How to be a quantitative ecologist. A survey of a coral reef uses 10 predefined linear transects covered by divers once every week. The response variable of interest is the abundance of a particular species of anemone as a function of water temperature. Counts of anemones are recorded at 20 regular line segments along the transect.\nLet\u0026rsquo;s simulate some data:\nset.seed(666) transects \u0026lt;- 10 data \u0026lt;- NULL for (tr in 1:transects){ # random effect (intercept) ref \u0026lt;- rnorm(1,0,.5) # water temperature gradient t \u0026lt;- runif(1, 18,22) + runif(1,-.2,0.2)*1:20 # Anemone gradient (expected response) ans \u0026lt;- exp(ref -14 + 1.8 * t - 0.045 * t^2) # actual counts on 20 segments of the current transect an \u0026lt;- rpois(20, ans) data \u0026lt;- rbind(data, cbind(rep(tr, 20), t, an)) } data \u0026lt;- data.frame(Transect = data[,1], Temperature = data[,2], Anemones = data[,3]) Standardize temperature:\nboo \u0026lt;- data$Temperature data$Temp \u0026lt;- (boo - mean(boo)) / sd(boo) head(data) Maximum-likelihood estimation With lme4, you write:\nfit_lme4 \u0026lt;- glmer(Anemones ~ Temp + I(Temp^2) + (1 | Transect), data = data, family = poisson) summary(fit_lme4) ## Generalized linear mixed model fit by maximum likelihood (Laplace ## Approximation) [glmerMod] ## Family: poisson ( log ) ## Formula: Anemones ~ Temp + I(Temp^2) + (1 | Transect) ## Data: data ## ## AIC BIC logLik deviance df.resid ## 1380.2 1393.4 -686.1 1372.2 196 ## ## Scaled residuals: ## Min 1Q Median 3Q Max ## -2.32356 -0.66154 -0.03391 0.63964 2.67283 ## ## Random effects: ## Groups Name Variance Std.Dev. ## Transect (Intercept) 0.1497 0.3869 ## Number of obs: 200, groups: Transect, 10 ## ## Fixed effects: ## Estimate Std. Error z value Pr(\u0026gt;|z|) ## (Intercept) 3.95436 0.12368 31.971 \u0026lt;2e-16 *** ## Temp -0.04608 0.02435 -1.892 0.0584 . ## I(Temp^2) -0.11122 0.01560 -7.130 1e-12 *** ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 ## ## Correlation of Fixed Effects: ## (Intr) Temp ## Temp 0.037 ## I(Temp^2) -0.118 -0.294 Bayesian analysis with Jags In Jags, we fit the corresponding GLMM with:\nmodel \u0026lt;- function() { for (i in 1:n){ count[i] ~ dpois(lambda[i]) log(lambda[i]) \u0026lt;- a[transect[i]] + b[1] * x[i] + b[2] * pow(x[i],2) } for (j in 1:nbtransects){ a[j] ~ dnorm (mu.a, tau.a) } mu.a ~ dnorm (0, 0.01) tau.a \u0026lt;- pow(sigma.a, -2) sigma.a ~ dunif (0, 100) b[1] ~ dnorm (0, 0.01) b[2] ~ dnorm (0, 0.01) } dat \u0026lt;- list(n = nrow(data), nbtransects = transects, x = data$Temp, count = data$Anemones, transect = data$Transect) inits \u0026lt;- function() list(a = rnorm(transects), b = rnorm(2), mu.a = rnorm(1), sigma.a = runif(1)) par \u0026lt;- c (\u0026quot;a\u0026quot;, \u0026quot;b\u0026quot;, \u0026quot;mu.a\u0026quot;, \u0026quot;sigma.a\u0026quot;) fit \u0026lt;- jags(data = dat, inits = inits, parameters.to.save = par, model.file = model, n.chains = 2, n.iter = 5000, n.burn = 1000) ## Compiling model graph ## Resolving undeclared variables ## Allocating nodes ## Graph information: ## Observed stochastic nodes: 200 ## Unobserved stochastic nodes: 14 ## Total graph size: 1622 ## ## Initializing model round(fit$BUGSoutput$summary[, -c(4,6)], 3) ## mean sd 2.5% 50% 97.5% Rhat n.eff ## a[1] 4.383 0.026 4.332 4.384 4.435 1.001 2000 ## a[2] 3.223 0.050 3.120 3.224 3.322 1.001 2000 ## a[3] 3.584 0.058 3.471 3.584 3.696 1.001 2000 ## a[4] 3.620 0.054 3.513 3.622 3.721 1.001 2000 ## a[5] 4.092 0.050 3.990 4.092 4.190 1.001 2000 ## a[6] 3.848 0.038 3.774 3.849 3.921 1.000 2000 ## a[7] 4.476 0.029 4.419 4.475 4.531 1.002 1300 ## a[8] 4.398 0.055 4.288 4.397 4.499 1.003 550 ## a[9] 3.835 0.033 3.771 3.835 3.900 1.001 1700 ## a[10] 4.076 0.029 4.018 4.076 4.132 1.001 2000 ## b[1] -0.047 0.024 -0.093 -0.047 0.000 1.001 2000 ## b[2] -0.111 0.016 -0.142 -0.111 -0.079 1.004 450 ## deviance 1324.697 4.651 1317.432 1324.185 1335.440 1.001 2000 ## mu.a 3.950 0.160 3.617 3.957 4.272 1.005 310 ## sigma.a 0.482 0.140 0.296 0.457 0.826 1.003 1000 Bayesian analysis with brms What about with brms?\nbayes.brms \u0026lt;- brm(Anemones ~ Temp + I(Temp^2) + (1 | Transect), data = data, family = poisson(\u0026quot;log\u0026quot;), chains = 2, # nb of chains iter = 5000, # nb of iterations, including burnin warmup = 1000, # burnin thin = 1) ## Running /Library/Frameworks/R.framework/Resources/bin/R CMD SHLIB foo.c ## clang -mmacosx-version-min=10.13 -I\u0026quot;/Library/Frameworks/R.framework/Resources/include\u0026quot; -DNDEBUG -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/Rcpp/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/unsupported\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/BH/include\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/src/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppParallel/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/rstan/include\u0026quot; -DEIGEN_NO_DEBUG -DBOOST_DISABLE_ASSERTS -DBOOST_PENDING_INTEGER_LOG2_HPP -DSTAN_THREADS -DBOOST_NO_AUTO_PTR -include '/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp' -D_REENTRANT -DRCPP_PARALLEL_USE_TBB=1 -I/usr/local/include -fPIC -Wall -g -O2 -c foo.c -o foo.o ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:88: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:1: error: unknown type name 'namespace' ## namespace Eigen { ## ^ ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:16: error: expected ';' after top level declarator ## namespace Eigen { ## ^ ## ; ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:96:10: fatal error: 'complex' file not found ## #include \u0026lt;complex\u0026gt; ## ^~~~~~~~~ ## 3 errors generated. ## make: *** [foo.o] Error 1 ## ## SAMPLING FOR MODEL '9326e0aa5a2dca4008c9e5b2ee4e86bf' NOW (CHAIN 1). ## Chain 1: ## Chain 1: Gradient evaluation took 4.5e-05 seconds ## Chain 1: 1000 transitions using 10 leapfrog steps per transition would take 0.45 seconds. ## Chain 1: Adjust your expectations accordingly! ## Chain 1: ## Chain 1: ## Chain 1: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 1: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 1: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 1: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 1: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 1: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 1: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 1: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 1: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 1: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 1: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 1: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 1: ## Chain 1: Elapsed Time: 0.893226 seconds (Warm-up) ## Chain 1: 4.91683 seconds (Sampling) ## Chain 1: 5.81006 seconds (Total) ## Chain 1: ## ## SAMPLING FOR MODEL '9326e0aa5a2dca4008c9e5b2ee4e86bf' NOW (CHAIN 2). ## Chain 2: ## Chain 2: Gradient evaluation took 2.4e-05 seconds ## Chain 2: 1000 transitions using 10 leapfrog steps per transition would take 0.24 seconds. ## Chain 2: Adjust your expectations accordingly! ## Chain 2: ## Chain 2: ## Chain 2: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 2: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 2: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 2: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 2: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 2: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 2: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 2: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 2: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 2: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 2: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 2: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 2: ## Chain 2: Elapsed Time: 1.21769 seconds (Warm-up) ## Chain 2: 4.45281 seconds (Sampling) ## Chain 2: 5.6705 seconds (Total) ## Chain 2: Display results:\nbayes.brms ## Family: poisson ## Links: mu = log ## Formula: Anemones ~ Temp + I(Temp^2) + (1 | Transect) ## Data: data (Number of observations: 200) ## Draws: 2 chains, each with iter = 5000; warmup = 1000; thin = 1; ## total post-warmup draws = 8000 ## ## Group-Level Effects: ## ~Transect (Number of levels: 10) ## Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS ## sd(Intercept) 0.47 0.14 0.29 0.81 1.00 1366 2089 ## ## Population-Level Effects: ## Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS ## Intercept 3.95 0.16 3.63 4.26 1.00 1359 1751 ## Temp -0.05 0.02 -0.09 0.00 1.00 4126 4490 ## ITempE2 -0.11 0.02 -0.14 -0.08 1.00 4018 4209 ## ## Draws were sampled using sampling(NUTS). For each parameter, Bulk_ESS ## and Tail_ESS are effective sample size measures, and Rhat is the potential ## scale reduction factor on split chains (at convergence, Rhat = 1). Visualize:\nplot(bayes.brms) We can assess the quality of fit of this model:\npp_check(bayes.brms, ndraws = 100, type = 'ecdf_overlay') As expected, the fit is almost perfect because we simulated data from this very model.\nWhat if we\u0026rsquo;d like to test the effect of temperature using WAIC?\nWe fit a model with no effect of temperature:\nbayes.brms2 \u0026lt;- brm(Anemones ~ 1 + (1 | Transect), data = data, family = poisson(\u0026quot;log\u0026quot;), chains = 2, # nb of chains iter = 5000, # nb of iterations, including burnin warmup = 1000, # burnin thin = 1) ## Running /Library/Frameworks/R.framework/Resources/bin/R CMD SHLIB foo.c ## clang -mmacosx-version-min=10.13 -I\u0026quot;/Library/Frameworks/R.framework/Resources/include\u0026quot; -DNDEBUG -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/Rcpp/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/unsupported\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/BH/include\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/src/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppParallel/include/\u0026quot; -I\u0026quot;/Library/Frameworks/R.framework/Versions/4.1/Resources/library/rstan/include\u0026quot; -DEIGEN_NO_DEBUG -DBOOST_DISABLE_ASSERTS -DBOOST_PENDING_INTEGER_LOG2_HPP -DSTAN_THREADS -DBOOST_NO_AUTO_PTR -include '/Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp' -D_REENTRANT -DRCPP_PARALLEL_USE_TBB=1 -I/usr/local/include -fPIC -Wall -g -O2 -c foo.c -o foo.o ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:88: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:1: error: unknown type name 'namespace' ## namespace Eigen { ## ^ ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/src/Core/util/Macros.h:628:16: error: expected ';' after top level declarator ## namespace Eigen { ## ^ ## ; ## In file included from \u0026lt;built-in\u0026gt;:1: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/StanHeaders/include/stan/math/prim/mat/fun/Eigen.hpp:13: ## In file included from /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Dense:1: ## /Library/Frameworks/R.framework/Versions/4.1/Resources/library/RcppEigen/include/Eigen/Core:96:10: fatal error: 'complex' file not found ## #include \u0026lt;complex\u0026gt; ## ^~~~~~~~~ ## 3 errors generated. ## make: *** [foo.o] Error 1 ## ## SAMPLING FOR MODEL 'fd9804091450c044661531e89123d1dc' NOW (CHAIN 1). ## Chain 1: ## Chain 1: Gradient evaluation took 4.4e-05 seconds ## Chain 1: 1000 transitions using 10 leapfrog steps per transition would take 0.44 seconds. ## Chain 1: Adjust your expectations accordingly! ## Chain 1: ## Chain 1: ## Chain 1: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 1: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 1: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 1: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 1: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 1: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 1: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 1: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 1: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 1: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 1: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 1: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 1: ## Chain 1: Elapsed Time: 0.704799 seconds (Warm-up) ## Chain 1: 2.58736 seconds (Sampling) ## Chain 1: 3.29216 seconds (Total) ## Chain 1: ## ## SAMPLING FOR MODEL 'fd9804091450c044661531e89123d1dc' NOW (CHAIN 2). ## Chain 2: ## Chain 2: Gradient evaluation took 1.8e-05 seconds ## Chain 2: 1000 transitions using 10 leapfrog steps per transition would take 0.18 seconds. ## Chain 2: Adjust your expectations accordingly! ## Chain 2: ## Chain 2: ## Chain 2: Iteration: 1 / 5000 [ 0%] (Warmup) ## Chain 2: Iteration: 500 / 5000 [ 10%] (Warmup) ## Chain 2: Iteration: 1000 / 5000 [ 20%] (Warmup) ## Chain 2: Iteration: 1001 / 5000 [ 20%] (Sampling) ## Chain 2: Iteration: 1500 / 5000 [ 30%] (Sampling) ## Chain 2: Iteration: 2000 / 5000 [ 40%] (Sampling) ## Chain 2: Iteration: 2500 / 5000 [ 50%] (Sampling) ## Chain 2: Iteration: 3000 / 5000 [ 60%] (Sampling) ## Chain 2: Iteration: 3500 / 5000 [ 70%] (Sampling) ## Chain 2: Iteration: 4000 / 5000 [ 80%] (Sampling) ## Chain 2: Iteration: 4500 / 5000 [ 90%] (Sampling) ## Chain 2: Iteration: 5000 / 5000 [100%] (Sampling) ## Chain 2: ## Chain 2: Elapsed Time: 0.648008 seconds (Warm-up) ## Chain 2: 2.85026 seconds (Sampling) ## Chain 2: 3.49826 seconds (Total) ## Chain 2: Then we compare both models, by ranking them with their WAIC or using ELPD:\n(waic1 \u0026lt;- waic(bayes.brms)) # waic model w/ tempterature ## ## Computed from 8000 by 200 log-likelihood matrix ## ## Estimate SE ## elpd_waic -668.2 9.8 ## p_waic 11.0 1.1 ## waic 1336.3 19.7 (waic2 \u0026lt;- waic(bayes.brms2)) # waic model wo/ tempterature ## ## Computed from 8000 by 200 log-likelihood matrix ## ## Estimate SE ## elpd_waic -702.4 13.3 ## p_waic 12.3 1.2 ## waic 1404.9 26.5 ## ## 2 (1.0%) p_waic estimates greater than 0.4. We recommend trying loo instead. loo_compare(waic1, waic2) ## elpd_diff se_diff ## bayes.brms 0.0 0.0 ## bayes.brms2 -34.3 8.9 Conclusions In all examples, you can check that i) there is little or no difference between Jags and brms numeric summaries of posterior distributions, and ii) maximum likelihood parameter estimates (glm(), lmer() and glmer() functions) and posterior means/medians are very close to each other.\nIn contrast to Jags, Nimble or Stan, you do not need to code the likelihood yourself in brms, as long as it is available in the package. You can have a look to a list of distributions with ?brms::brmsfamily, and there is also a possibility to define custom distributions with custom_family().\nNote I haven\u0026rsquo;t covered diagnostics of convergence, more here. You can also integrate brms outputs in a tidy workflow with tidybayes, and use the ggplot() magic.\nAs usual, the code is on GitHub, visit https://github.com/oliviergimenez/fit-glmm-with-brms.\nNow I guess it\u0026rsquo;s up to the students to pick their favorite Bayesian tool.\n","date":1650240000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1650240000,"objectID":"d5f7f86edeba830f6265ddb3e1c354aa","permalink":"https://oliviergimenez.github.io/blog/glmm-brms/","publishdate":"2022-04-18T00:00:00Z","relpermalink":"/blog/glmm-brms/","section":"blog","summary":"Here I illustrate how to fit GLMMs with the R package brms, and compare to Jags and lme4.\n","tags":["GLMM","bayes","R","brms","jags"],"title":"Bayesian analyses made easy: GLMMs in R package brms","type":"blog"},{"authors":[],"categories":[],"content":"8th conf of 🇫🇷 Statistical Ecology network #gdrEcoStat is about to start 🥳 We\u0026#39;re at 🏡 this year in @montpellier_ I\u0026#39;ll be tweeting today \u0026amp; tomorrow, hope it\u0026#39;ll pick your interest 😇🔇\n➡️ https://t.co/1MvnlFtSh3@INEE_CNRS @CNRS_OccitaniE @cefemontpellier @LbbeLyon @Ifremer_fr pic.twitter.com/PgeEPfx4UI\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 4, 2022 📸🥳 L\u0026#39;enregistrement vidéo des journées du #gdrEcoStat sont disponibles sur https://t.co/zptxN4JVol 😎 Jetez un coup d\u0026#39;oeil à la description de la vidéo pour accéder à une présentation par titre/orateur-trice. PDF dispo ici https://t.co/YmGutsYaQQ 😉\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 12, 2022 ","date":1649721600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1649721600,"objectID":"7ab945f6a0f716847106b5f254773ac8","permalink":"https://oliviergimenez.github.io/blog/2022-04-12-8emes-journees-du-gdr-ecostat/","publishdate":"2022-04-12T00:00:00Z","relpermalink":"/blog/2022-04-12-8emes-journees-du-gdr-ecostat/","section":"blog","summary":"8th conf of 🇫🇷 Statistical Ecology network #gdrEcoStat is about to start 🥳 We\u0026#39;re at 🏡 this year in @montpellier_ I\u0026#39;ll be tweeting today \u0026amp; tomorrow, hope it\u0026#39;ll pick your interest 😇🔇","tags":["statistical-ecology","conference"],"title":"8èmes journées du GdR EcoStat","type":"blog"},{"authors":[],"categories":[],"content":"📢🥳 With @MaudQueroue and @ValentinLauret we gave a short introduction to nimble @R_nimble nimble is a #rstats 📦 to fit models in the Bayesian framework w/ MCMC, it\u0026#39;s also a programming environment for using/coding fns/distns/samplers\n➡️ Check out https://t.co/MuUjGc4U8i pic.twitter.com/RBtu07v874\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 8, 2022 ","date":1649376000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1649376000,"objectID":"7945083d6b6a43f4ba7b6957aec6ed60","permalink":"https://oliviergimenez.github.io/blog/2022-04-08-introductory-workshop-on-nimble/","publishdate":"2022-04-08T00:00:00Z","relpermalink":"/blog/2022-04-08-introductory-workshop-on-nimble/","section":"blog","summary":"📢🥳 With @MaudQueroue and @ValentinLauret we gave a short introduction to nimble @R_nimble nimble is a #rstats 📦 to fit models in the Bayesian framework w/ MCMC, it\u0026#39;s also a programming environment for using/coding fns/distns/samplers","tags":["nimble","R","bayesian-statistics"],"title":"Introductory workshop on Nimble","type":"blog"},{"authors":[],"categories":[],"content":"A pleasure to work w/ #ComputoJournal @Computo85445972 for our paper on trade-offs bw #DeepLearning for species id \u0026amp; inference on predator-prey co-occurrence, which comes w/ a reproducible R workflow 😇https://t.co/Tgo6OJs7r0#OpenAccess #ReproducibleResearch #RStats 🧵⬇️ https://t.co/x2w73ezAcs\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 6, 2022 ","date":1649203200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1649203200,"objectID":"69b140c187d97f0d340b0806327cc72c","permalink":"https://oliviergimenez.github.io/blog/2022-04-06-new-paper-on-trade-offs-between-deep-learning-for-species-id-inference-on-predator-prey-co-occurrence/","publishdate":"2022-04-06T00:00:00Z","relpermalink":"/blog/2022-04-06-new-paper-on-trade-offs-between-deep-learning-for-species-id-inference-on-predator-prey-co-occurrence/","section":"blog","summary":"A pleasure to work w/ #ComputoJournal @Computo85445972 for our paper on trade-offs bw #DeepLearning for species id \u0026amp; inference on predator-prey co-occurrence, which comes w/ a reproducible R workflow 😇https://t.co/Tgo6OJs7r0#OpenAccess #ReproducibleResearch #RStats 🧵⬇️ https://t.","tags":["deep-learning","dynamic-occupancy","R"],"title":"New paper on trade-offs between Deep Learning for species id inference on predator-prey co-occurrence","type":"blog"},{"authors":[],"categories":[],"content":"Aaaaaaand it\u0026#39;s out 🤩🥳🐺🇫🇷 Awesome paper led by @oksanagrente \u0026quot;Wolf depredation hotspots in France: Clustering analyses adjusting for livestock availability\u0026quot; w/ @OFBiodiversite\nFree PDF ➡️ https://t.co/DilNeA8ikL@INEE_CNRS @CNRS_OccitaniE @umontpellier pic.twitter.com/ZcsbCANcw3\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 25, 2022 ","date":1645747200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1645747200,"objectID":"fdd691252bc768ca3dc3dfc5bba5be15","permalink":"https://oliviergimenez.github.io/blog/2022-02-25-new-paper-on-wolf-depredation-hotspots-in-france/","publishdate":"2022-02-25T00:00:00Z","relpermalink":"/blog/2022-02-25-new-paper-on-wolf-depredation-hotspots-in-france/","section":"blog","summary":"Aaaaaaand it\u0026#39;s out 🤩🥳🐺🇫🇷 Awesome paper led by @oksanagrente \u0026quot;Wolf depredation hotspots in France: Clustering analyses adjusting for livestock availability\u0026quot; w/ @OFBiodiversite\nFree PDF ➡️ https://t.co/DilNeA8ikL@INEE_CNRS @CNRS_OccitaniE @umontpellier pic.twitter.com/ZcsbCANcw3\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 25, 2022 ","tags":["loup","wolf"],"title":"New paper on wolf depredation hotspots in France","type":"blog"},{"authors":[],"categories":[],"content":"With advice by @fxcoudert and the help \u0026amp; patience of the @hal_fr folks, I ↗️ my #OpenAccess score 66% ➡️ 90% 🤩 https://t.co/etLjASx4fD I felt like a paleontologist digging up fossils, uploading the accepted or non-formated version of papers to public archives #OpenAccess https://t.co/y9dJhVUVE5 pic.twitter.com/ahKhNp0TgS\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 10, 2022 ","date":1644451200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1644451200,"objectID":"917a417f27827f83fbb25f460a5d8d90","permalink":"https://oliviergimenez.github.io/blog/2022-02-10-my-open-access-score/","publishdate":"2022-02-10T00:00:00Z","relpermalink":"/blog/2022-02-10-my-open-access-score/","section":"blog","summary":"With advice by @fxcoudert and the help \u0026amp; patience of the @hal_fr folks, I ↗️ my #OpenAccess score 66% ➡️ 90% 🤩 https://t.co/etLjASx4fD I felt like a paleontologist digging up fossils, uploading the accepted or non-formated version of papers to public archives #OpenAccess https://t.","tags":["open-access","open-science"],"title":"My Open Access score","type":"blog"},{"authors":[],"categories":[],"content":"Je fais une introduction aux statistiques bayésiennes avec R les 11-12 avril pour les doctorantes et doctorants @umontpellier du @CollegeDoctoUM 😇\nInscription ➡️ https://t.co/w5PFBGyHSz\nDiapos \u0026amp; vidéos ➡️ https://t.co/O5rAQTSksE#BayesianStatistics #rstats pic.twitter.com/VqIUcSLQja\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 1, 2022 ","date":1643673600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1643673600,"objectID":"b521f6143975524ce4bf96c4bf25ce1b","permalink":"https://oliviergimenez.github.io/blog/2022-02-01-introduction-aux-statistiques-bayesiennes-avec-r/","publishdate":"2022-02-01T00:00:00Z","relpermalink":"/blog/2022-02-01-introduction-aux-statistiques-bayesiennes-avec-r/","section":"blog","summary":"Je fais une introduction aux statistiques bayésiennes avec R les 11-12 avril pour les doctorantes et doctorants @umontpellier du @CollegeDoctoUM 😇\nInscription ➡️ https://t.co/w5PFBGyHSz\nDiapos \u0026amp; vidéos ➡️ https://t.co/O5rAQTSksE#BayesianStatistics #rstats pic.","tags":["statistiques-bayésiennes"],"title":"Introduction aux statistiques bayésiennes avec R","type":"blog"},{"authors":[],"categories":[],"content":"🤩😇 This year I am going to learn a lot working closely with amazing young researchers - I\u0026#39;m on ☁️9⃣ 🥳\nShort intro below 🧵 Give them a follow to stay up-to-date w/ their research (1/7)#ResearchIsLife @cefemontpellier @CNRS_OccitaniE @INEE_CNRS @umontpellier @twitthair1 pic.twitter.com/CBiL5fGvV0\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) January 15, 2022 ","date":1642204800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1642204800,"objectID":"fc4a69850abb477f243e3a12154730f1","permalink":"https://oliviergimenez.github.io/blog/2022-01-15-new-early-career-researchers-in-the-team/","publishdate":"2022-01-15T00:00:00Z","relpermalink":"/blog/2022-01-15-new-early-career-researchers-in-the-team/","section":"blog","summary":"🤩😇 This year I am going to learn a lot working closely with amazing young researchers - I\u0026#39;m on ☁️9⃣ 🥳\nShort intro below 🧵 Give them a follow to stay up-to-date w/ their research (1/7)#ResearchIsLife @cefemontpellier @CNRS_OccitaniE @INEE_CNRS @umontpellier @twitthair1 pic.","tags":["ECR","early-career-researchers"],"title":"New early career researchers in the team!","type":"blog"},{"authors":[],"categories":[],"content":"Faire une thèse est en général une expérience enrichissante pendant laquelle on apprend des tas de choses en recherche et sur soi-même. On interagit avec des gens géniaux. Mais c\u0026#39;est aussi trop souvent une expérience traumatisante. Il faut le dire. En parler. Et faire qqch 🧵1/14\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) January 8, 2022 ","date":1641600000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1641600000,"objectID":"e08fd4fff82613f8e52f87205a2fbb5d","permalink":"https://oliviergimenez.github.io/blog/2022-01-08-these-et-sante-mentale/","publishdate":"2022-01-08T00:00:00Z","relpermalink":"/blog/2022-01-08-these-et-sante-mentale/","section":"blog","summary":"Faire une thèse est en général une expérience enrichissante pendant laquelle on apprend des tas de choses en recherche et sur soi-même. On interagit avec des gens géniaux. Mais c\u0026#39;est aussi trop souvent une expérience traumatisante.","tags":["academic-life","mental-health"],"title":"Thèse et santé mentale","type":"blog"},{"authors":null,"categories":null,"content":"Here I illustrate how to train a CNN with Keras in R to predict from patients\u0026rsquo; CT scans those who will develop severe illness from Covid.\nMotivation Michael Blum tweeted about the STOIC2021 - COVID-19 AI challenge. The main goal of this challenge is to predict from the patients\u0026rsquo; CT scans who will develop severe illness from Covid.\nGiven my recent interest in machine learning, this challenge peaked my interest. Although Python is the machine learning lingua franca, it is possible to train a convolutional neural network (CNN) in R and perform (binary) image classification.\nHere, I will use an R interface to Keras that allows training neural networks. Note that the dataset shared for the challenge is big, like 280Go big, and it took me a day to download it. For the sake of illustration, I will use a similar but much lighter dataset from a Kaggle repository https://www.kaggle.com/plameneduardo/sarscov2-ctscan-dataset.\nThe code is available on GitHub as usual https://github.com/oliviergimenez/bin-image-classif.\nFirst things first, load the packages we will need.\nlibrary(tidyverse) theme_set(theme_light()) library(keras) Read in and process data We will need a function to process images, I\u0026rsquo;m stealing that one written by Spencer Palladino.\nprocess_pix \u0026lt;- function(lsf) { img \u0026lt;- lapply(lsf, image_load, grayscale = TRUE) # grayscale the image arr \u0026lt;- lapply(img, image_to_array) # turns it into an array arr_resized \u0026lt;- lapply(arr, image_array_resize, height = 100, width = 100) # resize arr_normalized \u0026lt;- normalize(arr_resized, axis = 1) #normalize to make small numbers return(arr_normalized) } Now let\u0026rsquo;s process images for patients with Covid, and do some reshaping. Idem with images for patients without Covid.\n# with covid lsf \u0026lt;- list.files(\u0026quot;dat/COVID/\u0026quot;, full.names = TRUE) covid \u0026lt;- process_pix(lsf) covid \u0026lt;- covid[,,,1] # get rid of last dim covid_reshaped \u0026lt;- array_reshape(covid, c(nrow(covid), 100*100)) # without covid lsf \u0026lt;- list.files(\u0026quot;dat/non-COVID/\u0026quot;, full.names = TRUE) ncovid \u0026lt;- process_pix(lsf) ncovid \u0026lt;- ncovid[,,,1] # get rid of last dim ncovid_reshaped \u0026lt;- array_reshape(ncovid, c(nrow(ncovid), 100*100)) We have 1252 CT scans of patients with Covid, and 1229 without.\nLet\u0026rsquo;s visualise these scans. Let\u0026rsquo;s pick a patient with Covid, and another one without.\nscancovid \u0026lt;- reshape2::melt(covid[10,,]) plotcovid \u0026lt;- scancovid %\u0026gt;% ggplot() + aes(x = Var1, y = Var2, fill = value) + geom_raster() + labs(x = NULL, y = NULL, title = \u0026quot;CT scan of a patient with covid\u0026quot;) + scale_fill_viridis_c() + theme(legend.position = \u0026quot;none\u0026quot;) scanncovid \u0026lt;- reshape2::melt(ncovid[10,,]) plotncovid \u0026lt;- scanncovid %\u0026gt;% ggplot() + aes(x = Var1, y = Var2, fill = value) + geom_raster() + labs(x = NULL, y = NULL, title = \u0026quot;CT scan of a patient without covid\u0026quot;) + scale_fill_viridis_c() + theme(legend.position = \u0026quot;none\u0026quot;) library(patchwork) plotcovid + plotncovid Put altogether and shuffle.\ndf \u0026lt;- rbind(cbind(covid_reshaped, 1), # 1 = covid cbind(ncovid_reshaped, 0)) # 0 = no covid set.seed(1234) shuffle \u0026lt;- sample(nrow(df), replace = F) df \u0026lt;- df[shuffle, ] Sounds great. We have everything we need to start training a convolutional neural network model.\nConvolutional neural network (CNN) Let\u0026rsquo;s build our training and testing datasets using a 80/20 split.\nset.seed(2022) split \u0026lt;- sample(2, nrow(df), replace = T, prob = c(0.8, 0.2)) train \u0026lt;- df[split == 1,] test \u0026lt;- df[split == 2,] train_target \u0026lt;- df[split == 1, 10001] # label in training dataset test_target \u0026lt;- df[split == 2, 10001] # label in testing dataset Now build our model. I use three layers (layer_dense() function) that I put one after the other with piping. I also use regularization (layer_dropout() function) to avoid overfitting.\nmodel \u0026lt;- keras_model_sequential() %\u0026gt;% layer_dense(units = 512, activation = \u0026quot;relu\u0026quot;) %\u0026gt;% layer_dropout(0.4) %\u0026gt;% layer_dense(units = 256, activation = \u0026quot;relu\u0026quot;) %\u0026gt;% layer_dropout(0.3) %\u0026gt;% layer_dense(units = 128, activation = \u0026quot;relu\u0026quot;) %\u0026gt;% layer_dropout(0.2) %\u0026gt;% layer_dense(units = 2, activation = 'softmax') Compile the model with defaults specific to binary classification.\nmodel %\u0026gt;% compile(optimizer = 'adam', loss = 'binary_crossentropy', metrics = c('accuracy')) We use one-hot encoding (to_categorical() function) aka dummy coding in statistics.\ntrain_label \u0026lt;- to_categorical(train_target) test_label \u0026lt;- to_categorical(test_target) Now let\u0026rsquo;s fit our model to the training dataset.\nfit_covid \u0026lt;- model %\u0026gt;% fit(x = train, y = train_label, epochs = 25, batch_size = 512, # try also 128 and 256 verbose = 2, validation_split = 0.2) A quick visualization of the performances shows that the algorithm is doing not too bad. No over/under-fitting. Accuracy and loss are fine.\nplot(fit_covid) What about the performances on the testing dataset?\nmodel %\u0026gt;% evaluate(test, test_label) ## loss accuracy ## 0.02048795 0.99795920 Let\u0026rsquo;s do some predictions on the testing dataset, and compare with ground truth.\npredictedclasses \u0026lt;- model %\u0026gt;% predict_classes(test) table(Prediction = predictedclasses, Actual = test_target) ## Actual ## Prediction 0 1 ## 0 243 0 ## 1 1 246 Pretty cool. Only one healthy patient is misclassified as being sick. Let\u0026rsquo;s save our model for further use.\nsave_model_tf(model, \u0026quot;model/covidmodel\u0026quot;) # save the model I\u0026rsquo;m happy with these results. In general however, we need to find ways to improve the performances. Check out some tips here with examples implemented in Keras with R there.\n","date":1641081600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1641081600,"objectID":"f099b0db0ad21cdc8aefcd61b2c9274d","permalink":"https://oliviergimenez.github.io/blog/image-classif/","publishdate":"2022-01-02T00:00:00Z","relpermalink":"/blog/image-classif/","section":"blog","summary":"Here I illustrate how to train a CNN with Keras in R to predict from patients\u0026rsquo; CT scans those who will develop severe illness from Covid.\n","tags":["deep learning","keras","R","rstats"],"title":"Binary image classification using Keras in R: Using CT scans to predict patients with Covid","type":"blog"},{"authors":[],"categories":[],"content":"Looking back to 2021, I don\u0026#39;t feel like emphasizing papers or projects. I guess this crazy period has reminded us of the importance of human relationships. Science is made by people, but people are not just scientists. Stay safe, and take care of others and yourself ⬇️🧵 1/11 pic.twitter.com/LUHPPnI1c0\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 21, 2021 ","date":1640044800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1640044800,"objectID":"3da43161105208b3692aa1504871327c","permalink":"https://oliviergimenez.github.io/blog/2021-12-21-looking-back-to-2021/","publishdate":"2021-12-21T00:00:00Z","relpermalink":"/blog/2021-12-21-looking-back-to-2021/","section":"blog","summary":"Looking back to 2021, I don\u0026#39;t feel like emphasizing papers or projects. I guess this crazy period has reminded us of the importance of human relationships. Science is made by people, but people are not just scientists.","tags":["retrospective"],"title":"Looking back to 2021","type":"blog"},{"authors":[],"categories":[],"content":"Under the kind influence of @FredBarraquand \u0026amp; @TiagoALOMarques, I have been thinking lately about #OpenAccess and #OpenScience. Access to science should be free. Period. My two cents. ⬇️🧵 1/11 #AcademicTwitter #AcademicChatter pic.twitter.com/n2F5p5qwvQ\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 19, 2021 ","date":1639872000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1639872000,"objectID":"499bb78830225dd1d43b6e9fabf219f3","permalink":"https://oliviergimenez.github.io/blog/2021-12-19-my-two-cents-on-open-science/","publishdate":"2021-12-19T00:00:00Z","relpermalink":"/blog/2021-12-19-my-two-cents-on-open-science/","section":"blog","summary":"Under the kind influence of @FredBarraquand \u0026amp; @TiagoALOMarques, I have been thinking lately about #OpenAccess and #OpenScience. Access to science should be free. Period. My two cents. ⬇️🧵 1/11 #AcademicTwitter #AcademicChatter pic.","tags":["open-science"],"title":"My two cents on Open Science","type":"blog"},{"authors":null,"categories":null,"content":"I have been interviewed recently by Lise Viollat who\u0026rsquo;s preparing a documentary for our national research group in statistical ecology. I have written my two cents answers to her challenging and very relevant questions. Thank you Lise for the interview !\nCan you introduce yourself? What is your job about? Your background? My work is motivated by questions on the ecology of large mammals, and large carnivores in particular (wolves, bears and lynx).\nWhat questions ? Very basic questions at first glance: How many animals? Where are they? How to explain trends in population abundance and species distribution?\nThese questions might seem simple, but the issue is that large carnivores are difficult to see in the field. We need methods to sample these populations, to manage uncertainties in counting, in detecting individuals and species, and to predict as well, where and how many. This is where statistical ecology comes in.\nI am what you might call an ecological statistician, and I work for the National Centre for Scientific Research.\nI have a rather standard background. I studied mathematics up to Bachelor\u0026rsquo;s degree. A degree that I repeated, I still had nightmares about it until recently. Then, from the Master\u0026rsquo;s degree onwards, I turned to applied mathematics and statistics. I embarked on a PhD in statistics applied to ecology, and to the dynamics of animal populations, with the estimation of survival and dispersal. As a post-doc, I spent time in England and Scotland to see what it was like elsewhere, to build my own research network, and to learn English.\nOnce recruited, I became increasingly interested in evolutionary ecology and conservation issues. I also worked more and more on issues related to large mammals, and species that interact with human activities.\nTo better understand these interactions, we need the human and social sciences. So I went back to studying sociology. I took a few courses, but I couldn\u0026rsquo;t go all the way to the Master\u0026rsquo;s degree. I am still learning on the job.\nWhat is ecology for you? I am certainly not the right person to answer this question, I never studied ecology at university. I learned from my students and colleagues, and on the job.\nI like definitions, so I\u0026rsquo;m going for it. Ecology is the study of the processes that influence the distribution and abundance of species, the interactions between these species, and the interactions between these species and their environment. Maud Quéroué\u0026rsquo;s thesis is an example of the study of these interactions between species using new statistical models.\nIt\u0026rsquo;s the science of interactions in a nutshell, and that\u0026rsquo;s what I like about this discipline. Between species, between species and their environment. By species, I also mean humans. And here we come back to this idea of mobilising ecology, human and social sciences, and the study of socio-ecosystems in which we mix ecology, politics, economics, sociology and anthropology. Gilles Maurer\u0026rsquo;s thesis is an illustration of this approach, and combines demography, genetics, economics and anthropology to study the relationship between Asian elephants and humans.\nFinally, when we talk about ecology, we must understand it in a broad sense, with the idea that we are also talking about the evolutionary processes that contribute to explaining variability in biological systems. The theses of Marlène Gamelon and Mathieu Buoro are examples of work combining ecology and evolution to better understand the dynamics of animal populations.\nHow would you define statistical ecology? And modelling? In statistical ecology, statistical is the adjective that qualifies the noun ecology. It is ecology that motivates, that asks the questions. This does not make statistics a mere toolbox, but you have to accept the statements of ecology to make relevant statistics.\nComing from mathematics and statistics, this is something that did not come naturally to me. In statistics, one often sees the data from ecology as the numerical application to show that the statistical method works in the real world (often in addition to simulated data).\nI learned the importance of motivating statistics by ecological questions from Jean-Dominique Lebreton, one of my thesis directors, who himself got it from Jean-Marie Legay who thought a lot about the place of modelling in biology, and interdisciplinarity.\nStatistical ecology therefore means applying and above all developing statistical methods and tools to answer questions in ecology. In France, the scientific community meets on these themes in the statistical ecology research group which we created in 2014.\nStatistics in general is the science of collecting, analysing and drawing conclusions from data. All three are important for ecology. How we collect data to test ecological hypotheses. How we analyse it. And how you interpret the results.\nIt is at the analysis stage that models are mostly used. A model in this context is an abstraction of reality, based on what is known, and is used to extract information using probabilities. An example of a statistical model is a simple linear regression in which we try to pass a line through a scatter plot to explain a trend.\nHow is statistical ecology useful? Why is it important to ask the questions \u0026ldquo;where are the species?\u0026rdquo;, and \u0026ldquo;how many animals are there?\u0026rdquo;? The species we study are in contact with humans and their activities. Fishing activities for bottlenose dolphins in the Mediterranean, for example, or sheep farming for wolves in France, to take another example.\nIn this context, I would say that statistical ecology and modelling serve three main purposes. To understand, to predict, and to help in decision making.\nI\u0026rsquo;ll take the example of the wolf, on which we do a lot of work with colleagues from the French Office for Biodiversity.\nFirst of all, we need to understand. We need to know the factors that explain the distribution of the species, this is information that allows us to establish the conservation status of a species. It was Julie Louvrier\u0026rsquo;s thesis that showed that the species is plastic and how it has recolonised the country.\nPredicting then. Julie also showed that it was possible to predict the short-term dispersal of the species with a mathematical model that was fitted to the data. As another example, Nina Santostasi in her thesis evaluated the effectiveness of different management scenarios for hybridisation between dogs and wolves in Italy.\nFinally, informing the management decision process. The species is managed by the French state, and it is on the basis of the numbers we produce thanks to statistical ecology that the maximum number of culls is established each year. This was Lucile Marescot\u0026rsquo;s thesis. We also need to know the effect of shooting on wolf predation of domestic livestock, a complex problem in spatial statistics. This is Oksana Grente\u0026rsquo;s recently defended thesis.\nTo sum up, there is a lot of discussion about biodiversity and ecosystem services, and therefore about the interdependence between humans and nature. A connection that is increasingly forgotten, to quote a theme dear to Alix Cosquer who is part of our team and works in environmental psychology.\nIn this context, statistical ecology is used to understand, predict and inform decision-making on the interactions between species, animals and humans, and their environment.\nWhy did you choose this career? Several reasons, in no particular order.\nIt is an incredible privilege to be paid to be a researcher, especially as a civil servant. Personally, I did not want to experience the difficult working conditions of my parents. Professionnaly, this status offers good conditions for doing our job. A permanent position allows us to work as a collective. This is an important point because research is an adventure involving several people, a team adventure; you don\u0026rsquo;t achieve anything, or not much, when you work alone (at least for me). You don\u0026rsquo;t do good research by pushing for more competition.\nOur status also allows us to take a long view, a long view that is essential for quality research, research that also serves society. Our work on the wolf, which I mentioned, began more than 10 years ago! This long time frame is being undermined by the precarization of employment which is hitting the research world hard, with more and more fixed-term contracts, and I am worried about the consequences for the younger generation of course. More competition. More administration. More excellence. Go fast. Quantity. Productivity. Fewer jobs. Fewer resources. Research and teaching are managed like a company, we have to optimise, rationalise, students are customers. It can\u0026rsquo;t work. Sorry, I\u0026rsquo;m going astray\u0026hellip; Another reason I chose this career is that I wanted to tackle applied problems, which could have concrete applications for people, for society. Or at least to contribute a little bit in that sense. To be useful, basically.\nAnother reason is that it is very (very) rewarding to accompany and support the new generation of scientists. It\u0026rsquo;s exciting, I love to walk the path with those who are embarking on a master\u0026rsquo;s degree, a thesis or a post-doc.\nFinally, this job allows you to be on the move all the time in your head. You learn all the time. You are constantly intellectually stimulated. I love that.\nDo you have to be good at math to do this job? We all have a maths hump. The sense of number, of counting. It is by training this sense that we improve, that we become better at maths. Like with a musical instrument.\nThe main obstacle is surely the language, because maths is also a language. Again like with music. You have to learn it to read it, to write it, and to understand what people who do it say to each other. With some efforts, you can really enjoy yourself.\nWhat\u0026rsquo;s important to realise is that you can use maths in ecology at any level. From the simple rule of three, which I have mastered to the use of stochastic differential equations, which I have not understood. Everyone can find what they are looking for, and do relevant research. For statistics I recommend David Spiegelhalter\u0026rsquo;s book which introduces the discipline in a very accessible way.\nWhat do you like about what you do? First of all, it\u0026rsquo;s the people I work with. My biggest motivation comes from my students. I want to work with them, to exchange, to interact. It\u0026rsquo;s all about interactions. It\u0026rsquo;s also the people in our team. I\u0026rsquo;m learning all the time. And it\u0026rsquo;s the people in the statistical ecology community in France and abroad. It\u0026rsquo;s a community that doesn\u0026rsquo;t take itself too seriously, but that does serious and exciting research to answer the big questions about the biodiversity crisis. It is also very nice to feel part of a community. By the way, we are organising the next GDR Ecologie Statistique days in Montpellier at the beginning of April 2022, don\u0026rsquo;t hesitate to join us!\nThe second thing, as I have already mentioned, is the fact that you are always on the move intellectually. You learn, you try things out, you explore, you play, you write, you read, you think, all the while feeling useful at your level in the construction of knowledge.\nThe third thing is sharing. Sharing knowledge by accompanying the students first of all, as I mentioned. And teaching too. I don\u0026rsquo;t do much as I\u0026rsquo;m already having trouble keeping my head above water with my job as a researcher. I admire my colleagues at universities who do two jobs, they are researchers and teachers! In sharing, I also include workshops that we offer over several days during which we train folks in statistical ecology methods. These workshops are free and video recorded so that as many people as possible have access to them.\nWhat are the future challenges of ecology? I have no idea! I don\u0026rsquo;t really consider myself an ecologist. The only things that come to mind are related to our research topics, of course. I think about biodiversity conservation. How do we find a balance, and a sustainable balance, between our activities and the nature around us? This means understanding better the interactions between human and non-human actors, why these interactions can lead to conflicts and how to resolve them. It also means working on complex systems, the so-called socio-ecosystems, and therefore opening up ecology even more to the human and social sciences.\nAnd the future challenges for statistics and modelling? I\u0026rsquo;m not necessarily more comfortable talking about future challenges for statistics and modelling, but we\u0026rsquo;ve thought about it a bit with colleagues from the GDR EcoStat, so I\u0026rsquo;m a bit less dry than for ecology. I see three.\nFirstly, the modelling of socio-ecosystems. There is a need to integrate different disciplines in agent-based approaches (economics, geography, ethnology). The question is how to use modelling approaches to estimate the many quantities in these complex models.\nWe have an explosion of technological means deployed at all scales of organisation. I am thinking of camera traps, environmental DNA, GPS, drones, aerial or satellite photos, etc. We are faced with a huge amount of data from which it is often difficult to extract the relevant information. The second challenge is therefore that of massive data, and the development of approaches that allow us to learn from this data and automate tasks that become impossible for humans to do. I am obviously thinking of machine learning, and deep learning in particular.\nThe third issue is predictive ecology. How can we predict ecological phenomena? This requires the manipulation of large-scale systems that couple ecology and physics, for example, or the ocean and the atmosphere, to take another example. It also requires thinking about how to propagate uncertainties throughout the analysis, right up to the prediction.\nIs there a tool, a school of thought, a discovery that has changed everything for you in your field? Yes, clearly! It is the introduction of Bayesian theory in the early 2000s. I discovered this theory as a post-doc in England and then in Scotland with my colleague Ruth King and others.\nWhat is the Bayesian theory? The idea is to update the a priori knowledge we have of a biological system with the data we accumulate on this system. This a priori knowledge can be seen as the confidence we have in a hypothesis, and the Bayesian theory allows us to update this degree of confidence thanks to the data. The Bayesian theory allows us to formalise the natural way of reasoning, to induce the chances that a hypothesis is true or false.\nThe Bayesian theory is used everywhere in science, but also in the everyday world, for example in medicine to make a diagnosis, in computer science to calibrate the anti-spam of your e-mail, in finance to manage stock portfolios, and it even allowed Alan Turing to crack the Enigma machine which allowed the Germans to communicate by coded messages.\nThe strength of the Bayesian theory also comes from the fact that it comes with very powerful algorithms to put it into practice. Very powerful in the sense that we can use increasingly complex models that are closer and closer to what we know about the species we are studying, models that we could not dream of using before.\nTo give you a concrete example, I can think of one. In ecology, we often try to model the distribution of species. This requires data on large areas to be sure that we have sampled all the habitats that are favourable or unfavourable to the species. This is difficult to do with a demanding and expensive scientific protocol. We can see that participatory science makes it possible to sample large areas at lower cost. But the data are biased because if we see a lot where there are observers, that doesn\u0026rsquo;t mean that where there is no one, the species isn\u0026rsquo;t there; it simply means that where there are people, the species is detected. Bayesian analysis makes it possible to combine these different types of monitoring to quantify distribution by combining the advantages of both types of monitoring, unbiased data from scientific protocols, and wide spatial coverage from participatory programmes. This is the thesis of Valentin Lauret.\nDon\u0026rsquo;t get me started on Bayesian theory, I teach it in our Master\u0026rsquo;s programme and to PhD students in our doctoral school and I can\u0026rsquo;t get enough of it!\nCould you give an example of your work? Recently, we\u0026rsquo;ve been working a lot on the lynx in France. The main cause of mortality for the lynx is collisions with cars. A major challenge is therefore to find ways of reducing this mortality by collision. We have developed a model that allows us to evaluate scenarios to reduce this mortality. For example, building a wildlife crossing, basically a tunnel or a bridge through which animals can cross the road without risk. We wanted this tool to be used by those involved in the lynx dossier, and in particular by the regional government. We therefore encapsulated all the modelling components in a software that can be used by non-specialists. Everything is done in R, so it is open and reproducible!\nWhy did I choose to talk about this project specifically?\nFirstly because it is a project that mixes different types of modelling, statistical and computer science, different disciplines with ecology and statistics, and sub-disciplines of ecology with landscape ecology, movement ecology, and population dynamics.\nSecondly, because it was a project in which I questioned myself a lot. We chose to show the modelling to the actors with workshops where we presented the models and where we were put in place. The stakeholders had very high expectations, both in terms of the models, their use, the outputs, the long-term maintenance of the software and the updating of the data. In the end, co-construction enabled us to be much more relevant than if we had done the same modelling work in the laboratory.\nAlso in human terms, it is a project that ended well, since the post-doct who was the link between the different partners and who did most of the work, Sarah Bauduin, was recruited by the French Office for Biodiversity.\nFinally, it is a project that opens up other research questions, with a focus on connectivity problems due to fragmentation by land transport infrastructures (car and train). This is Maëlis Kervellec\u0026rsquo;s thesis which she has just started.\nAny message for young people who might be interested in this route? You mean the old fart\u0026rsquo;s message to the younger generation? I always find it difficult to give a message, to give advice. Especially because for me it worked, but clearly I was lucky, very lucky. What worked for me will not necessarily work for others. In that sense my message and my advice are not worth much.\nI\u0026rsquo;m going to cite Ted Lasso, and I\u0026rsquo;m going to say: be interested, read, ask questions, be curious about what other people are doing, and why they are doing it. Don\u0026rsquo;t listen too much to the advice of old people, and make your own way.\nAnd have fun, don\u0026rsquo;t take yourself too seriously!\nAn anecdote to share? A rather positive consequence of confinements. It\u0026rsquo;s the story of a training course in population dynamics that we offered in 2006 with about 30 participants at that time. We\u0026rsquo;re going for a reboot next year with two colleagues and friends of the team, Aurélien Besnard and Sarah Cubaynes. One tweet and a few messages to some mailing lists later, we have 2315 subscribers after 15 days! We\u0026rsquo;re not sure how we\u0026rsquo;re going to manage all these people, but it\u0026rsquo;s very exciting! Many of the people who have signed up come from countries where access to knowledge is difficult for many reasons. So we\u0026rsquo;re doing a free, online course, and we\u0026rsquo;re recording it. That\u0026rsquo;s where I\u0026rsquo;d like to go, more knowledge sharing. It sounds a bit grandiloquent, but there is a great need for biodiversity conservation.\n📢👋 Together with @SarahCubaynes and @abesnardEPHE we will be giving a 2-day introductory workshop March 21-22 on quantitative methods for population dynamics in R #rstats Join us 🥳 It\u0026#39;s online and free of charge 😉 You just need to register https://t.co/I5sNS5kDOo Please RT 😇 pic.twitter.com/rkw8JWZBJi\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) October 11, 2021 ","date":1638748800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1638748800,"objectID":"a283cd9c5c3b79896cc376d8c98c96e8","permalink":"https://oliviergimenez.github.io/blog/interview-ecostat-english/","publishdate":"2021-12-06T00:00:00Z","relpermalink":"/blog/interview-ecostat-english/","section":"blog","summary":"I have been interviewed recently by Lise Viollat who\u0026rsquo;s preparing a documentary for our national research group in statistical ecology. I have written my two cents answers to her challenging and very relevant questions. Thank you Lise for the interview !\n","tags":["statistical ecology","interview"],"title":"Why a career in statistical ecology?","type":"blog"},{"authors":null,"categories":null,"content":"Dans le cadre d\u0026rsquo;un documentaire que réalise Lise Viollat, celle-ci m\u0026rsquo;a posé des questions sur mon parcours en recherche, et en écologie statistique. Je reproduis mes réponses ici. Merci Lise pour l\u0026rsquo;entretien !\nEst-ce que vous pouvez vous présenter? En quoi consiste votre travail, votre parcours ? Mon travail est motivé par des questions sur l’écologie des grands mammifères, et les grands carnivores en particulier (loups, ours et lynx).\nQuelles questions ? Très basiquement : combien y a-t-il d’animaux, et où sont-ils, et pourquoi ces effectifs et cette distribution.\nCes questions peuvent paraître très simples, mais le gros problème est qu’on ne peut pas facilement voir ces animaux. Il faut des méthodes pour échantillonner, pour gérer les incertitudes de comptage, de détection des individus et des espèces, et pour prédire aussi, combien et où. Et c’est là que les statistiques entrent en jeu.\nJe travaille en écologie statistique, je suis ce qu’on pourrait appeler un éco-statistien. Et je travaille au CNRS, le Centre National de la Recherche Scientifique.\nPour arriver au CNRS, j’ai eu un parcours plutôt classique. J’ai étudié les mathématiques jusqu’à la licence. Une licence que j’ai redoublée, j’en faisais encore des cauchemars jusqu’à récemment. Puis à partir du Master je me suis orienté vers les mathématiques appliquées avec les statistiques. Je me suis embarqué dans une thèse en statistique appliquée à l’écologie, et à la dynamique des populations d\u0026rsquo;animaux, avec l’estimation de la survie et de la dispersion. Après la thèse à Montpellier, j\u0026rsquo;ai fait un détour par Grenoble, puis je suis parti en Angleterre et en Ecosse pour voir comment c’était ailleurs, me faire mon propre réseau, et surtout apprendre l’anglais.\nUne fois recruté, je me suis de plus en plus intéressé aux questions d’écologie évolutive, et de conservation. J’ai aussi travaillé de plus en plus sur des questions en lien avec les grands mammifères, des espèces en interaction avec les activités humaines.\nPour mieux comprendre ces interactions, on fait appel aux sciences humaines et sociales. J’ai donc repris des études en sociologie. J\u0026rsquo;ai suivi quelques cours en licence, mais je n’ai pas pu aller au bout du Master. Je continue de me former sur le tas.\nQu\u0026rsquo;est-ce que pour vous l\u0026rsquo;écologie ? Je ne suis certainement pas la bonne personne pour répondre à la question, je n’ai jamais étudié l’écologie à l’université. J’ai appris au contact de mes étudiantes, de mes étudiants, et de mes collègues, et sur le tas.\nJ’aime bien les définitions, alors je me lance. L’écologie c’est l’étude des processus qui influencent la distribution et l’abondance des espèces, les interactions entre ces espèces, et les interactions entre ces espèces et leur environnement. La thèse de Maud Quéroué est un exemple de l\u0026rsquo;étude de ces interactions entre espèces grâce à de nouveaux modèles statistiques.\nC’est la science des interactions en bref, et c’est ça que j’aime dans cette discipline. Entre les espèces, entre les espèces et leur environnement. Dans les espèces, je mets aussi les humains. Et pour le coup on revient à cette idée de mobiliser écologie et sciences humaines et sociales, et l’étude des socio-écosystèmes dans lesquels on mélange écologie, politique, économie, sociologie et anthropologie. La thèse de Gilles Maurer est une illustration de cette démarche, et combine démographie, génétique, économie et anthropologie pour étudier les relations entre éléphant d’Asie et humains.\nPour finir, quand on parle d’écologie, il faut l’entendre au sens large, avec l’idée qu’on parle aussi des processus évolutifs qui contribuent à expliquer la variabilité dans les systèmes biologiques. Les thèses de Marlène Gamelon et Mathieu Buoro sont des exemples de travaux mêlant écologie et évolution pour mieux comprendre la dynamique des populations animales.\nComment est-ce que vous définiriez l\u0026rsquo;écologie statistique ? La modélisation ? Dans écologie statistique, l’ordre des mots est important. On ne dit pas statistique pour l’écologie. C’est l’écologie qui motive, qui pose les questions. Les statistiques n\u0026rsquo;en deviennent pas pour autant une simple boîte à outils, mais il faut accepter les énoncés de l’écologie pour faire des statistiques pertinentes.\nVenant des mathématiques et des statistiques, c’est quelque chose qui ne m’était pas naturel. En statistiques, on voit souvent les données de l’écologie comme l’application numérique pour montrer que la méthode statistique marche avec un exemple de données réelles (en complément aux données simulées souvent).\nJ’ai appris l’importance de motiver les statistiques par les questions écologiques au contact de Jean-Dominique Lebreton un de mes directeurs de thèse, qui le tenait lui-même de Jean-Marie Legay qui a beaucoup réfléchi sur la place de la modélisation en biologie, et l’interdisciplinarité.\nL’écologie statistique c’est donc appliquer et surtout développer des méthodes et des outils statistiques pour répondre aux questions de l’écologie. En France, la communauté scientifique se retrouve sur ces thèmes dans le groupement de recherche en écologie statistique que nous avons créé en 2014.\nLa statistique de manière générale, c’est la science qui permet de collecter, analyser et tirer des conclusions à partir des données. Les trois sont importants pour l’écologie. Comment on récolte des données pour tester des hypothèses écologiques. Comment on les analyse. Et comment on interprète les résultats.\nC’est à l’étape d’analyse qu’on utilise surtout des modèles. Un modèle dans ce contexte c’est une abstraction de la réalité, qui se base sur ce qu’on sait, et qu’on utilise pour extraire de l’information en utilisant les probabilités. Un exemple de modèle statistique c’est la régression linéaire simple dans laquelle on essaye de faire passer une droite dans un nuage de point pour expliquer une tendance.\nA quoi elles servent d\u0026rsquo;après vous ? Pourquoi est-il important de poser les questions \u0026ldquo;où sont les espèces?\u0026rdquo;, et \u0026ldquo;combien y a-t-il d’animaux?\u0026rdquo;. Les espèces qu’on étudie sont au contact des humains, et de leurs activités. Les activités de pêche pour le grand dauphin en Méditerranée par exemple, ou l’élévage d’ovins pour les loups en France pour prendre un autre exemple.\nDans ce contexte, je dirais que l’écologie statistique et la modélisation servent principalement à trois choses. Comprendre, prédire, et aider à la décision.\nJe prends l’exemple du loup sur lequel on travaille beaucoup avec les collègues de l’ Office Français de la Biodiversité.\nComprendre d’abord. On a besoin de connaître les facteurs qui expliquent la distribution de l’espèce, ce sont des informations qui permettent d’établir le statut de conservation par exemple, celui de l’UICN. C’était la thèse de Julie Louvrier qui a permis de montrer que l’espèce est plastique et la façon dont l’espèce a recolonisé le pays.\nPrédire ensuite. Julie a aussi montré qu’on pouvait prédire à court terme la dispersion de l’espèce avec un modèle mathématique qu’on confrontait aux données. Autre exemple, dans sa thèse, Nina Santostasi a évalué l\u0026rsquo;efficacité de différents scénarios de gestion de l\u0026rsquo;hybridation entre chiens et loups en Italie.\nEnfin, l’aide à la décision. L’espèce est gérée par l’état français, et c’est entre autres sur la base d’effectifs que nous produisons grâce à l’écologie statistique que le nombre maximal de prélèvements est établi chaque année. C’était la thèse de Lucile Marescot. On a aussi besoin de connaître l’effet des tirs mis en place sur la prédation par le loup des troupeaux domestiques, un problème complexe de statistique spatiale. C’est la thèse d’Oksana Grente qui a soutenu récemment.\nPour résumer, on parle beaucoup de biodiversité et de services rendus par les écosystèmes, et donc d’interdépendance entre les humains et la nature. Une connexion qui s’oublie de plus en plus pour reprendre un thème cher à Alix Cosquer qui travaille dans notre équipe en psychologie de l’environnement.\nDans ce contexte, l’écologie statistique sert à comprendre, à prédire et à informer la prise de décision sur les interactions entre les espèces, animaux et humains, et leur environnement.\nPourquoi avoir choisi cette carrière ? Plusieurs raisons, dans le désordre.\nC’est un privilège incroyable que d’être payé pour faire le métier de chercheur, qui plus est avec le statut de fonctionnaire. A titre personnel, je ne voulais pas connaître les conditions de travail difficiles de mes parents. A titre professionnel, ce statut offre de bonnes conditions pour faire notre métier. Un poste permanent permet de travailler en collaborant avec les collègues. C’est un point important car la recherche c’est une aventure à plusieurs, une aventure d’équipe ; on n’arrive à rien, ou à pas grand-chose, quand on travaille seul (en tous cas moi). On ne fait pas une bonne recherche en poussant à plus de compétition. Notre statut permet aussi de s’inscrire dans un temps long, ce temps long qui est indispensable à une recherche de qualité, une recherche qui sert la société aussi. Nos travaux sur le loup dont je parlais ont commencé il y a plus de 10 ans ! Ce temps long est mis à mal par la précarisation de l’emploi qui touche le monde de la recherche de plein fouet, avec de plus en plus de CDDs, et je m’inquiète des conséquences pour la jeune génération bien sûr. Plus de compétition. Plus d’administratif, plus d’excellence. Aller vite. Quantité. Productivité. Moins de postes. Moins de moyens. On gère la recherche et l’enseignement comme une entreprise, il faut optimiser, rationnaliser, les étudiants sont des clients. Ca ne peut pas marcher. Désolé je m\u0026rsquo;égare\u0026hellip;\nUne autre raison pour laquelle j\u0026rsquo;ai choisi cette carrière est que j’avais envie de m’attaquer à des problèmes appliqués, qui pourraient avoir des applications concrètes pour les gens, pour la société. Ou au moins de contribuer un peu en ce sens. Être utile en gros.\nUn autre raison est qu\u0026rsquo;il est très (très) gratifiant d’accompagner la nouvelle génération de scientifiques. C’est excitant, j’adore faire un bout de chemin avec celles et ceux qui s\u0026rsquo;embarquent dans un master, une thèse, ou un post-doc.\nEnfin, ce métier permet d’être tout le temps en mouvement dans sa tête. On apprend tout le temps. On est en permanence stimulé intellectuellement. J’adore ça.\nFaut-il être douer en math pour faire ce métier ? Ca veut dire quoi « doué en maths » en fait ? Ca fait penser à cette expression « avoir la bosse des maths ». Cette expression vient du 19ème siècle et d’une pseudo-science qui s’appelait la phrénologie. On établissait un lien entre vos capacités ou traits de caractères et les régions du cerveau, avec l’idée que ça se répercutait sur la forme du crâne. Les bons en maths avaient une bosse sur le front qui correspondait à l’endroit supposé du cerveau où se faisait les maths. C\u0026rsquo;est absurde! On a montré depuis que les mathématiques mobilisent différentes zones du cerveau.\nIl y a un bouquin très chouette de Stanislas Dehaene professeur au collège de France en psychochologie cognitive qui s’appelle justement la bosse des maths. Que dit ce bouquin? On a tous la bosse des maths. Le sens du nombre, du dénombrement. C’est en entrainant ce sens qu’on s’améliore, qu’on devient meilleur en maths. Comme avec un instrument de musique.\nLe principal obstacle est sûrement le langage, car les maths c’est aussi un langage. A nouveau comme en musique. Ca demande de l’apprendre pour la lire, l’écrire, et comprendre ce que se disent les gens qui en font. Une fois qu’on a fait nos gammes, on peut vraiment s’amuser, se faire plaisir.\nCe qui est important de réaliser, c’est qu’on peut utiliser les maths en écologie quel que soit son niveau. De la simple règle de trois que je maitrise à peu près, jusqu’à l’utilisation des équations différentielles stochastiques auxquelles je n’ai rien compris. Chacun et chacune peut y trouver son bonheur, et faire une recherche tout à fait pertinente. Pour les statistiques je conseille le livre de David Spiegelhalter qui introduit la discipline de façon très abordable.\nQu\u0026rsquo;est ce qui vous plait dans ce que vous faite ? C\u0026rsquo;est d\u0026rsquo;abord les personnes avec qui je travaille. Ma plus grosse motivation vient de mes étudiantes et de mes étudiants. J’ai envie de travailler avec eux, d’échanger, d’interagir. On en revient aux interactions. Ce sont aussi les gens de mon équipe. J’apprends tout le temps. Et ce sont les gens de la communauté en écologie statistique en France et à l’étranger. C’est une communauté qui ne se prend pas trop au sérieux, mais qui fait une recherche sérieuse et excitante pour répondre aux grandes questions qu’on se pose sur la crise de la biodiversité. C’est aussi très agréable de se sentir appartenir à une communauté. A ce propos, nous organisons les prochaines journées du GDR Ecologie Statistique à Montpellier début avril 2022, n\u0026rsquo;hésitez pas à nous rejoindre !\nLa deuxième chose, je l’ai déjà mentionné, c’est le fait d’exercer un métier dans lequel on est toujours en mouvement intellectuellement. On apprend, on essaye, on tente des choses, on explore, on joue, on écrit, on lit, on réfléchit, tout en se sentant utile à son niveau à la construction des connaissances.\nLa troisième chose c’est le partage. Le partage des connaissances via l’accompagnement des étudiantes et des étudiants d’abord, j’en ai parlé. Et l’enseignement aussi. J’en fais peu quand on compare aux collègues de l’enseignement supérieur qui forcent le respect. J’ai déjà du mal à garder la tête hors de l’eau avec mon métier de chercheur, eux en font deux, ils sont chercheurs et enseignants ! Dans le partage, je mets aussi des ateliers que nous proposons sur plusieurs jours au cours desquels on forme aux méthodes de l’écologie statistique. Ce sont des formations gratuites et filmées pour que le plus grand nombre y ait accès.\nQuels sont les futurs enjeux de l\u0026rsquo;écologie ? Je n\u0026rsquo;en sais rien du tout! Je ne me considère pas vraiment comme un écologue. Les seules choses qui me viennent à l’esprit sont en lien avec nos thématiques de recherche évidemment. Je pense à la conservation de la biodiversité. Comment trouver un équilibre, et un équilibre durable entre nos activités et la nature qui nous entoure? Ca sous-entend mieux comprendre les interactions entre les acteurs humains et non-humains, pourquoi ces interactions peuvent mener à des conflits et comment les résoudre. Ca veut dire aussi travailler sur des systèmes complexes, ces fameux socio-écosystèmes, et donc encore plus d’ouverture de l’écologie aux sciences humaines et sociales.\nDes statistiques et de la modélisation ? Je suis pas forcément plus à l’aise pour parler des enjeux en statistiques, mais on y a un peu réfléchi avec les collègues du GDR EcoStat, et je suis donc un peu moins sec que pour l’écologie. J’en vois trois.\nLa modélisation des socio-écosystèmes d’abord. Il y a besoin d’intégrer différentes disciplines dans des approches multi-agents (économie, géographie, ethnologie). On se demande comment utiliser les approches de modélisation pour estimer les quantités importantes de ces modèles complexes.\nOn a une explosion des moyens technologiques déployés à toutes les échelles d’organisation. Je pense aux pièges photos, à l\u0026rsquo;ADN environnemental, aux GPS, aux drones, aux photos aériennes ou satellites, etc. Nous sommes face à une masse conséquente de données dont il est souvent difficile d’extraire l’information pertinente. Le deuxième enjeu est donc celui des données massives, et du développement d’approches qui permettent d’apprendre de ces données, et d’automatiser des tâches qui deviennent impossibles à faire par des humains. Je pense évidemment au machine learning, et au deep learning en particulier.\nLe troisième enjeu est celui de l’écologie prédictive. Comment prédire les phénomènes écologiques? Ca demande de manipuler des systèmes de grandes dimensions qui couplent écologie et physique par exemple, ou océan et atmosphère pour prendre un autre exemple. Ca demande aussi de réfléchir à comment propager les incertitudes d’estimation et d’échantillonnage tout le long de l’analyse, jusqu’à la prédiction.\nY a-t-il un outil, un courant de pensée, une découverte qui pour vous a tout changé dans votre domaine ? Oui, clairement! C’est l’introduction de la théorie bayésienne au début des années 2000. J’ai découvert cette théorie en post-doc en Angleterre puis en Ecosse avec ma collègue Ruth King et d’autres.\nQu’est-ce que le bayésien ? L’idée est de mettre à jour la connaissance a priori qu’on a d’un système qu’on étudie grâce à l’expérimentation, avec les données qu’on accumule sur ce système. Cette connaissance a priori, on peut la voir comme la confiance qu’on a en une hypothèse, et le bayésien permet alors de mettre à jour ce degré de confiance grâce aux données. Le bayésien permet de formaliser la façon naturelle qu’on a de raisonner, d’induire les chances qu’une hypothèse soit vraie ou fausse.\nLa théorie bayésienne est utilisée un peu partout en sciences, mais aussi dans le monde de tous les jours par exemple en médecine pour faire un diagnostic, en informatique pour calibrer l’anti-spam de votre boîte électronique, en finance pour gérer des portfolios d’actions, et elle a même permis à Alan Turing de craquer la machine Enigma qui permettait aux allemands de communiquer par messages codés.\nLa force de la théorie bayésienne vient aussi du fait qu’elle vient avec des algorithmes pour la mettre en pratique qui sont très puissants. Très puissants au sens où on peut utiliser des modèles de plus en plus complexes et de plus en plus proches de ce qu’on sait des espèces qu’on étudie, des modèles qu’on ne pouvait pas rêver d’utiliser avant.\nPour vous donner du concret, je pense à un exemple. En écologie on cherche souvent à modéliser la distribution des espèces. Ca demande d’avoir des données sur des grandes surfaces pour être sûr qu’on a échantillonné tous les habitats favorables ou pas à l’espèce. C’est difficile à faire avec un protocole scientifique exigeant et coûteux. On voit que les sciences participatives permettent d’échantillonner à moindre coût sur de grandes surfaces. Mais les données sont biaisées car si on voit beaucoup là où il y a des observateurs et observatrices, ça ne veut pas dire que là où il n’y a personne, il n’y a pas l’espèce ; ça veut simplement dire que là où on a des gens, l’espèce est détectée. L’analyse bayésienne permet de combiner ces différents suivis pour quantifier la distribution en combinant les avantages des deux types de suivi, des données non-biaisées par les protocoles scientifiques, et une grande couverture spatiale par les programmes participatifs. C\u0026rsquo;est la thèse de Valentin Lauret.\nIl ne faut pas me lancer sur la théorie bayésienne, je l\u0026rsquo;enseigne en Master et pour les doctorants et doctorantes de notre école doctorale et je ne m\u0026rsquo;en lasse pas!\nPourriez vous donnez un exemple d\u0026rsquo;un de vos travaux ? C’est un travail sur le lynx en France. La première cause de mortalité du lynx sont les collisions avec les voitures. Un gros enjeu consiste donc à trouver des moyens de réduire cette mortalité par collision. On a développé un modèle qui permet d’évaluer des scénarios pour réduire cette mortalité. Par exemple construire un passage à faune, en gros un tunnel ou un pont par lequel les animaux peuvent traverser la route sans risque. On a voulu que cet outil puisse être utilisé par les acteurs du dossier lynx, et en particulier par l’état en région via les DREAL (Direction régionale de l\u0026rsquo;Environnement, de l\u0026rsquo;Aménagement et du Logement). On a donc encapsulé toutes les composantes modélisation dans un logiciel utilisable par des non-spécialistes. Tout est fait dans le logiciel R, donc le tout est ouvert et reproductible !\nPourquoi j’ai choisi ce projet ?\nD\u0026rsquo;abord parce que c\u0026rsquo;est un projet qui mélange différents types de modélisation, statistique et informatique, différentes disciplines avec écologie et statistique, et sous-disciplines de l’écologie avec écologie du paysage, écologie du déplacement, et la dynamique des populations.\nEnsuite parce qu\u0026rsquo;humainement c’est un projet au cours duquel je me suis beaucoup remis en question. On a choisi de montrer la modélisation aux acteurs avec des ateliers où on a présenté les modèles, et où on s’est fait remettre en place. Les acteurs avaient un niveau d’exigence très élevé, à la fois sur les modèles, sur leur utilisation, les sorties, le maintien à long terme du logiciel et la mise à jour des données. Au final, la co-construction nous a permis d’être beaucoup plus pertinents que si on avait fait le même travail de modélisation au laboratoire.\nToujours humainement, c’est un projet qui finit bien, puisque la post-doctorante qui a été le trait d’union entre les différents partenaires, et qui a fait l’essentiel du travail, Sarah Bauduin, a été recrutée à l’Office Français de la Biodiversité.\nEnfin, c’est un projet qui ouvre sur d’autres questions de recherche, avec un focus sur les problèmes de connectivité dus à la fragmentation par les infrastructures de transports terrestres (voiture et train). C’est la thèse de Maëlis Kervellec qu’elle vient tout juste de démarrer.\nUn message pour les jeunes qui pourrait être intéresser par cette voie ? Vous voulez dire le message du vieux con à la jeune génération? Je trouve toujours difficile de passer un message, de donner un conseil. D’autant que pour moi ça a marché, mais clairement j’ai eu de la chance, beaucoup de chance. Ce qui a marché pour moi ne marchera pas forcément pour d’autres. En ce sens mon message et mes conseils ne valent pas grand-chose.\nJe vais vous faire du Ted Lasso, et je dirais : intéressez-vous, lisez, posez des questions, soyez curieux de ce que font les autres, et pourquoi ils le font. N’écoutez pas trop les conseils des vieux, et faites votre propre chemin.\nEt rigolez, ne vous prenez pas trop au sérieux !\nUne anecdote à partager ? Une conséquence plutôt positive des confinements. C’est l’histoire d’une formation en dynamique des populations que nous avons proposée en 2006 avec une 30aine de participants à l’époque. J’ai relancé ça avec deux collègues et amis de l’équipe, Aurélien Besnard et Sarah Cubaynes. Un tweet et deux messages à des mailing lists plus tard, on en est à 2315 inscrits au bout de 15 jours ! On ne sait pas trop comment on va gérer tous ces gens, mais c’est très excitant ! Beaucoup de personnes inscrites viennent de pays où l\u0026rsquo;accès à la connaissance est difficile pour des tas de raisons. On fait donc une formation gratuite et en ligne, et on l’enregistre. C’est vers là que j’aimerais aller, plus de partage de la connaissance. Ça fait un peu grandiloquent, mais les besoins sont criants pour la conservation de la biodiversité.\n📢👋 Together with @SarahCubaynes and @abesnardEPHE we will be giving a 2-day introductory workshop March 21-22 on quantitative methods for population dynamics in R #rstats Join us 🥳 It\u0026#39;s online and free of charge 😉 You just need to register https://t.co/I5sNS5kDOo Please RT 😇 pic.twitter.com/rkw8JWZBJi\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) October 11, 2021 ","date":1638144000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1638144000,"objectID":"1c237477f79151fbe0b06632e7aaadf1","permalink":"https://oliviergimenez.github.io/blog/interview-ecostat/","publishdate":"2021-11-29T00:00:00Z","relpermalink":"/blog/interview-ecostat/","section":"blog","summary":"Dans le cadre d\u0026rsquo;un documentaire que réalise Lise Viollat, celle-ci m\u0026rsquo;a posé des questions sur mon parcours en recherche, et en écologie statistique. Je reproduis mes réponses ici. Merci Lise pour l\u0026rsquo;entretien !\n","tags":["statistical ecology","interview"],"title":"Pourquoi une carrière en écologie statistique?","type":"blog"},{"authors":null,"categories":null,"content":"I have a draft chapter on Bayes stats and MCMC at https://oliviergimenez.github.io/banana-book/crashcourse.html I\u0026rsquo;d love your feedback about what is confusing and what is missing 😇 #rstats\n","date":1632096000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1632096000,"objectID":"428b3e5a1631184039963f0cf1afb04d","permalink":"https://oliviergimenez.github.io/blog/draft-bayesmcmc/","publishdate":"2021-09-20T00:00:00Z","relpermalink":"/blog/draft-bayesmcmc/","section":"blog","summary":"I have a draft chapter on Bayes stats and MCMC at https://oliviergimenez.github.io/banana-book/crashcourse.html I\u0026rsquo;d love your feedback about what is confusing and what is missing 😇 #rstats\n","tags":["capture-recapture","Bayesian statistics","R","rstats"],"title":"Draft chapter on Bayes stats and MCMC in R","type":"blog"},{"authors":null,"categories":null,"content":"So, I\u0026rsquo;m writing a book 🤯 It\u0026rsquo;'\u0026rsquo;s called \u0026ldquo;Bayesian analysis of capture-recapture data with hidden Markov models - Theory and case studies in R\u0026rdquo;.\nSo, I\u0026#39;m writing a book 🤯 It\u0026#39;s called \u0026#39;Bayesian analysis of capture-recapture data w hidden Markov models - Theory and case studies in R\u0026#39;. Online version here https://t.co/Crgx5NX70s to be published w @CRCPress Will share chapter drafts 😇 \u0026amp; random thoughts along the way #rstats pic.twitter.com/ms6dIjvY7m\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) September 19, 2021 ","date":1632009600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1632009600,"objectID":"6df991e86f32cb717c3ca61c8d3c1342","permalink":"https://oliviergimenez.github.io/blog/banana-book/","publishdate":"2021-09-19T00:00:00Z","relpermalink":"/blog/banana-book/","section":"blog","summary":"So, I\u0026rsquo;m writing a book 🤯 It\u0026rsquo;'\u0026rsquo;s called \u0026ldquo;Bayesian analysis of capture-recapture data with hidden Markov models - Theory and case studies in R\u0026rdquo;.\n","tags":["capture-recapture","Bayesian statistics","R","rstats"],"title":"Bayesian analysis of capture-recapture data with hidden Markov models - Theory and case studies in R","type":"blog"},{"authors":null,"categories":null,"content":"I would like to familiarize myself with machine learning (ML) techniques in R. So I have been reading and learning by doing. I thought I\u0026rsquo;d share my experience for others who\u0026rsquo;d like to give it a try1.\nFirst version August 13, 2021, updated August 23, 2021 Since my first post, I’ve been reading notebooks shared by folks who ranked high in the challenge, and added two features that they used. Eventually, these new predictors did not help (I must be doing something wrong). I also explored some other ML algorithms. Last, I tuned the parameters more efficiently with a clever grid-search algorithm. All in all, I slightly improved my score, but most importantly, I now have a clean template for further use.\nMotivation All material available from GitHub at https://github.com/oliviergimenez/learning-machine-learning.\nThe two great books I’m using are:\nAn Introduction to Statistical Learning with Applications in R by Gareth James, Daniela Witten, Trevor Hastie and Robert Tibshirani\nTidy models in R by Max Kuhn and Julia Silge\nI also recommend checking out the material (codes, screencasts) shared by David Robinson and Julia Silge from whom I picked some useful tricks that I put to use below.\nTo try things, I’ve joined the Kaggle online community which gathers folks with lots of experience in ML from whom you can learn. Kaggle also hosts public datasets that can be used for playing around.\nI use the tidymodels metapackage that contains a suite of packages for modeling and machine learning using tidyverse principles. Check out all possibilities here, and parsnip models in particular there.\nLet’s start with the famous Titanic dataset. We need to predict if a passenger survived the sinking of the Titanic (1) or not (0). A dataset is provided for training our models (train.csv). Another dataset is provided (test.csv) for which we do not know the answer. We will predict survival for each passenger, submit our answer to Kaggle and see how well we did compared to other folks. The metric for comparison is the percentage of passengers we correctly predict – aka as accuracy.\nFirst things first, let’s load some packages to get us started.\nlibrary(tidymodels) # metapackage for ML library(tidyverse) # metapackage for data manipulation and visulaisation library(stacks) # stack ML models for better perfomance theme_set(theme_light()) doParallel::registerDoParallel(cores = 4) # parallel computations Data Read in training data.\nrawdata \u0026lt;- read_csv(\u0026quot;dat/titanic/train.csv\u0026quot;) glimpse(rawdata) ## Rows: 891 ## Columns: 12 ## $ PassengerId \u0026lt;dbl\u0026gt; 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20… ## $ Survived \u0026lt;dbl\u0026gt; 0, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, … ## $ Pclass \u0026lt;dbl\u0026gt; 3, 1, 3, 1, 3, 3, 1, 3, 3, 2, 3, 1, 3, 3, 3, 2, 3, 2, 3, 3, 2, 2, 3, … ## $ Name \u0026lt;chr\u0026gt; \u0026quot;Braund, Mr. Owen Harris\u0026quot;, \u0026quot;Cumings, Mrs. John Bradley (Florence Brig… ## $ Sex \u0026lt;chr\u0026gt; \u0026quot;male\u0026quot;, \u0026quot;female\u0026quot;, \u0026quot;female\u0026quot;, \u0026quot;female\u0026quot;, \u0026quot;male\u0026quot;, \u0026quot;male\u0026quot;, \u0026quot;male\u0026quot;, \u0026quot;male\u0026quot;,… ## $ Age \u0026lt;dbl\u0026gt; 22, 38, 26, 35, 35, NA, 54, 2, 27, 14, 4, 58, 20, 39, 14, 55, 2, NA, … ## $ SibSp \u0026lt;dbl\u0026gt; 1, 1, 0, 1, 0, 0, 0, 3, 0, 1, 1, 0, 0, 1, 0, 0, 4, 0, 1, 0, 0, 0, 0, … ## $ Parch \u0026lt;dbl\u0026gt; 0, 0, 0, 0, 0, 0, 0, 1, 2, 0, 1, 0, 0, 5, 0, 0, 1, 0, 0, 0, 0, 0, 0, … ## $ Ticket \u0026lt;chr\u0026gt; \u0026quot;A/5 21171\u0026quot;, \u0026quot;PC 17599\u0026quot;, \u0026quot;STON/O2. 3101282\u0026quot;, \u0026quot;113803\u0026quot;, \u0026quot;373450\u0026quot;, \u0026quot;330… ## $ Fare \u0026lt;dbl\u0026gt; 7.2500, 71.2833, 7.9250, 53.1000, 8.0500, 8.4583, 51.8625, 21.0750, 1… ## $ Cabin \u0026lt;chr\u0026gt; NA, \u0026quot;C85\u0026quot;, NA, \u0026quot;C123\u0026quot;, NA, NA, \u0026quot;E46\u0026quot;, NA, NA, NA, \u0026quot;G6\u0026quot;, \u0026quot;C103\u0026quot;, NA, N… ## $ Embarked \u0026lt;chr\u0026gt; \u0026quot;S\u0026quot;, \u0026quot;C\u0026quot;, \u0026quot;S\u0026quot;, \u0026quot;S\u0026quot;, \u0026quot;S\u0026quot;, \u0026quot;Q\u0026quot;, \u0026quot;S\u0026quot;, \u0026quot;S\u0026quot;, \u0026quot;S\u0026quot;, \u0026quot;C\u0026quot;, \u0026quot;S\u0026quot;, \u0026quot;S\u0026quot;, \u0026quot;S\u0026quot;, \u0026quot;S\u0026quot;,… naniar::miss_var_summary(rawdata) ## # A tibble: 12 × 3 ## variable n_miss pct_miss ## \u0026lt;chr\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; ## 1 Cabin 687 77.1 ## 2 Age 177 19.9 ## 3 Embarked 2 0.224 ## 4 PassengerId 0 0 ## 5 Survived 0 0 ## 6 Pclass 0 0 ## 7 Name 0 0 ## 8 Sex 0 0 ## 9 SibSp 0 0 ## 10 Parch 0 0 ## 11 Ticket 0 0 ## 12 Fare 0 0 After some data exploration (not shown), I decided to take care of missing values, gather the two family variables in a single variable, and create a variable title.\n# Get most frequent port of embarkation uniqx \u0026lt;- unique(na.omit(rawdata$Embarked)) mode_embarked \u0026lt;- as.character(fct_drop(uniqx[which.max(tabulate(match(rawdata$Embarked, uniqx)))])) # Build function for data cleaning and handling NAs process_data \u0026lt;- function(tbl){ tbl %\u0026gt;% mutate(class = case_when(Pclass == 1 ~ \u0026quot;first\u0026quot;, Pclass == 2 ~ \u0026quot;second\u0026quot;, Pclass == 3 ~ \u0026quot;third\u0026quot;), class = as_factor(class), gender = factor(Sex), fare = Fare, age = Age, ticket = Ticket, alone = if_else(SibSp + Parch == 0, \u0026quot;yes\u0026quot;, \u0026quot;no\u0026quot;), # alone variable alone = as_factor(alone), port = factor(Embarked), # rename embarked as port title = str_extract(Name, \u0026quot;[A-Za-z]+\\\\.\u0026quot;), # title variable title = fct_lump(title, 4)) %\u0026gt;% # keep only most frequent levels of title mutate(port = ifelse(is.na(port), mode_embarked, port), # deal w/ NAs in port (replace by mode) port = as_factor(port)) %\u0026gt;% group_by(title) %\u0026gt;% mutate(median_age_title = median(age, na.rm = T)) %\u0026gt;% ungroup() %\u0026gt;% mutate(age = if_else(is.na(age), median_age_title, age)) %\u0026gt;% # deal w/ NAs in age (replace by median in title) mutate(ticketfreq = ave(1:nrow(.), FUN = length), fareadjusted = fare / ticketfreq) %\u0026gt;% mutate(familyage = SibSp + Parch + 1 + age/70) } # Process the data dataset \u0026lt;- rawdata %\u0026gt;% process_data() %\u0026gt;% mutate(survived = as_factor(if_else(Survived == 1, \u0026quot;yes\u0026quot;, \u0026quot;no\u0026quot;))) %\u0026gt;% mutate(survived = relevel(survived, ref = \u0026quot;yes\u0026quot;)) %\u0026gt;% # first event is survived = yes select(survived, class, gender, age, alone, port, title, fareadjusted, familyage) # Have a look again glimpse(dataset) ## Rows: 891 ## Columns: 9 ## $ survived \u0026lt;fct\u0026gt; no, yes, yes, yes, no, no, no, no, yes, yes, yes, yes, no, no, no, y… ## $ class \u0026lt;fct\u0026gt; third, first, third, first, third, third, first, third, third, secon… ## $ gender \u0026lt;fct\u0026gt; male, female, female, female, male, male, male, male, female, female… ## $ age \u0026lt;dbl\u0026gt; 22, 38, 26, 35, 35, 30, 54, 2, 27, 14, 4, 58, 20, 39, 14, 55, 2, 30,… ## $ alone \u0026lt;fct\u0026gt; no, no, yes, no, yes, yes, yes, no, no, no, no, yes, yes, no, yes, y… ## $ port \u0026lt;fct\u0026gt; 3, 1, 3, 3, 3, 2, 3, 3, 3, 1, 3, 3, 3, 3, 3, 3, 2, 3, 3, 1, 3, 3, 2,… ## $ title \u0026lt;fct\u0026gt; Mr., Mrs., Miss., Mrs., Mr., Mr., Mr., Master., Mrs., Mrs., Miss., M… ## $ fareadjusted \u0026lt;dbl\u0026gt; 0.008136925, 0.080003704, 0.008894501, 0.059595960, 0.009034792, 0.0… ## $ familyage \u0026lt;dbl\u0026gt; 2.314286, 2.542857, 1.371429, 2.500000, 1.500000, 1.428571, 1.771429… naniar::miss_var_summary(dataset) ## # A tibble: 9 × 3 ## variable n_miss pct_miss ## \u0026lt;chr\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; ## 1 survived 0 0 ## 2 class 0 0 ## 3 gender 0 0 ## 4 age 0 0 ## 5 alone 0 0 ## 6 port 0 0 ## 7 title 0 0 ## 8 fareadjusted 0 0 ## 9 familyage 0 0 Let’s apply the same treatment to the test dataset.\nrawdata \u0026lt;- read_csv(\u0026quot;dat/titanic/test.csv\u0026quot;) holdout \u0026lt;- rawdata %\u0026gt;% process_data() %\u0026gt;% select(PassengerId, class, gender, age, alone, port, title, fareadjusted, familyage) glimpse(holdout) ## Rows: 418 ## Columns: 9 ## $ PassengerId \u0026lt;dbl\u0026gt; 892, 893, 894, 895, 896, 897, 898, 899, 900, 901, 902, 903, 904, 905… ## $ class \u0026lt;fct\u0026gt; third, third, second, third, third, third, third, second, third, thi… ## $ gender \u0026lt;fct\u0026gt; male, female, male, male, female, male, female, male, female, male, … ## $ age \u0026lt;dbl\u0026gt; 34.5, 47.0, 62.0, 27.0, 22.0, 14.0, 30.0, 26.0, 18.0, 21.0, 28.5, 46… ## $ alone \u0026lt;fct\u0026gt; yes, no, yes, yes, no, yes, yes, no, yes, no, yes, yes, no, no, no, … ## $ port \u0026lt;fct\u0026gt; 2, 3, 2, 3, 3, 3, 2, 3, 1, 3, 3, 3, 3, 3, 3, 1, 2, 1, 3, 1, 1, 3, 3,… ## $ title \u0026lt;fct\u0026gt; Mr., Mrs., Mr., Mr., Mrs., Mr., Miss., Mr., Mrs., Mr., Mr., Mr., Mrs… ## $ fareadjusted \u0026lt;dbl\u0026gt; 0.018730144, 0.016746411, 0.023175837, 0.020723684, 0.029395933, 0.0… ## $ familyage \u0026lt;dbl\u0026gt; 1.492857, 2.671429, 1.885714, 1.385714, 3.314286, 1.200000, 1.428571… naniar::miss_var_summary(holdout) ## # A tibble: 9 × 3 ## variable n_miss pct_miss ## \u0026lt;chr\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; ## 1 fareadjusted 1 0.239 ## 2 PassengerId 0 0 ## 3 class 0 0 ## 4 gender 0 0 ## 5 age 0 0 ## 6 alone 0 0 ## 7 port 0 0 ## 8 title 0 0 ## 9 familyage 0 0 Exploratory data analysis skimr::skim(dataset) Data summary Name dataset Number of rows 891 Number of columns 9 _______________________ Column type frequency: factor 6 numeric 3 ________________________ Group variables None Data summary\nVariable type: factor\nskim_variable n_missing complete_rate ordered n_unique top_counts survived 0 1 FALSE 2 no: 549, yes: 342 class 0 1 FALSE 3 thi: 491, fir: 216, sec: 184 gender 0 1 FALSE 2 mal: 577, fem: 314 alone 0 1 FALSE 2 yes: 537, no: 354 port 0 1 FALSE 4 3: 644, 1: 168, 2: 77, S: 2 title 0 1 FALSE 5 Mr.: 517, Mis: 182, Mrs: 125, Mas: 40 Variable type: numeric\nskim_variable n_missing complete_rate mean sd p0 p25 p50 p75 p100 hist age 0 1 29.39 13.26 0.42 21.00 30.00 35.00 80.00 ▂▇▃▁▁ fareadjusted 0 1 0.04 0.06 0.00 0.01 0.02 0.03 0.58 ▇▁▁▁▁ familyage 0 1 2.32 1.57 1.07 1.41 1.57 2.62 11.43 ▇▁▁▁▁ Let’s explore the data.\ndataset %\u0026gt;% count(survived) ## # A tibble: 2 × 2 ## survived n ## \u0026lt;fct\u0026gt; \u0026lt;int\u0026gt; ## 1 yes 342 ## 2 no 549 dataset %\u0026gt;% group_by(gender) %\u0026gt;% summarize(n = n(), n_surv = sum(survived == \u0026quot;yes\u0026quot;), pct_surv = n_surv / n) ## # A tibble: 2 × 4 ## gender n n_surv pct_surv ## \u0026lt;fct\u0026gt; \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; ## 1 female 314 233 0.742 ## 2 male 577 109 0.189 dataset %\u0026gt;% group_by(title) %\u0026gt;% summarize(n = n(), n_surv = sum(survived == \u0026quot;yes\u0026quot;), pct_surv = n_surv / n) %\u0026gt;% arrange(desc(pct_surv)) ## # A tibble: 5 × 4 ## title n n_surv pct_surv ## \u0026lt;fct\u0026gt; \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; ## 1 Mrs. 125 99 0.792 ## 2 Miss. 182 127 0.698 ## 3 Master. 40 23 0.575 ## 4 Other 27 12 0.444 ## 5 Mr. 517 81 0.157 dataset %\u0026gt;% group_by(class, gender) %\u0026gt;% summarize(n = n(), n_surv = sum(survived == \u0026quot;yes\u0026quot;), pct_surv = n_surv / n) %\u0026gt;% arrange(desc(pct_surv)) ## # A tibble: 6 × 5 ## # Groups: class [3] ## class gender n n_surv pct_surv ## \u0026lt;fct\u0026gt; \u0026lt;fct\u0026gt; \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; ## 1 first female 94 91 0.968 ## 2 second female 76 70 0.921 ## 3 third female 144 72 0.5 ## 4 first male 122 45 0.369 ## 5 second male 108 17 0.157 ## 6 third male 347 47 0.135 Some informative graphs.\ndataset %\u0026gt;% group_by(class, gender) %\u0026gt;% summarize(n = n(), n_surv = sum(survived == \u0026quot;yes\u0026quot;), pct_surv = n_surv / n) %\u0026gt;% mutate(class = fct_reorder(class, pct_surv)) %\u0026gt;% ggplot(aes(pct_surv, class, fill = class, color = class)) + geom_col(position = position_dodge()) + scale_x_continuous(labels = percent) + labs(x = \u0026quot;% in category that survived\u0026quot;, fill = NULL, color = NULL, y = NULL) + facet_wrap(~gender) dataset %\u0026gt;% mutate(age = cut(age, breaks = c(0, 20, 40, 60, 80))) %\u0026gt;% group_by(age, gender) %\u0026gt;% summarize(n = n(), n_surv = sum(survived == \u0026quot;yes\u0026quot;), pct_surv = n_surv / n) %\u0026gt;% mutate(age = fct_reorder(age, pct_surv)) %\u0026gt;% ggplot(aes(pct_surv, age, fill = age, color = age)) + geom_col(position = position_dodge()) + scale_x_continuous(labels = percent) + labs(x = \u0026quot;% in category that survived\u0026quot;, fill = NULL, color = NULL, y = NULL) + facet_wrap(~gender) dataset %\u0026gt;% ggplot(aes(fareadjusted, group = survived, color = survived, fill = survived)) + geom_histogram(alpha = .4, position = position_dodge()) + labs(x = \u0026quot;fare\u0026quot;, y = NULL, color = \u0026quot;survived?\u0026quot;, fill = \u0026quot;survived?\u0026quot;) dataset %\u0026gt;% ggplot(aes(familyage, group = survived, color = survived, fill = survived)) + geom_histogram(alpha = .4, position = position_dodge()) + labs(x = \u0026quot;family aged\u0026quot;, y = NULL, color = \u0026quot;survived?\u0026quot;, fill = \u0026quot;survived?\u0026quot;) Training/testing datasets Split our dataset in two, one dataset for training and the other one for testing. We will use an additionnal splitting step for cross-validation.\nset.seed(2021) spl \u0026lt;- initial_split(dataset, strata = \u0026quot;survived\u0026quot;) train \u0026lt;- training(spl) test \u0026lt;- testing(spl) train_5fold \u0026lt;- train %\u0026gt;% vfold_cv(5) Gradient boosting algorithms - xgboost Let’s start with gradient boosting methods which are very popular in the ML community.\nTuning Set up defaults.\nmset \u0026lt;- metric_set(accuracy) # metric is accuracy control \u0026lt;- control_grid(save_workflow = TRUE, save_pred = TRUE, extract = extract_model) # grid for tuning First a recipe.\nxg_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median step_dummy(all_nominal_predictors()) # all factors var are split into binary terms (factor disj coding) Then specify a gradient boosting model.\nxg_model \u0026lt;- boost_tree(mode = \u0026quot;classification\u0026quot;, # binary response trees = tune(), mtry = tune(), tree_depth = tune(), learn_rate = tune(), loss_reduction = tune(), min_n = tune()) # parameters to be tuned Now set our workflow.\nxg_wf \u0026lt;- workflow() %\u0026gt;% add_model(xg_model) %\u0026gt;% add_recipe(xg_rec) Use cross-validation to evaluate our model with different param config.\nxg_tune \u0026lt;- xg_wf %\u0026gt;% tune_grid(train_5fold, metrics = mset, control = control, grid = crossing(trees = 1000, mtry = c(3, 5, 8), # finalize(mtry(), train) tree_depth = c(5, 10, 15), learn_rate = c(0.01, 0.005), loss_reduction = c(0.01, 0.1, 1), min_n = c(2, 10, 25))) Visualize the results.\nautoplot(xg_tune) + theme_light() Collect metrics.\nxg_tune %\u0026gt;% collect_metrics() %\u0026gt;% arrange(desc(mean)) ## # A tibble: 162 × 12 ## mtry trees min_n tree_depth learn_rate loss_reduction .metric .estimator mean n ## \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;int\u0026gt; ## 1 3 1000 2 5 0.01 0.01 accuracy binary 0.849 5 ## 2 8 1000 2 5 0.01 0.01 accuracy binary 0.847 5 ## 3 8 1000 2 5 0.01 0.1 accuracy binary 0.846 5 ## 4 3 1000 2 15 0.01 0.1 accuracy binary 0.844 5 ## 5 5 1000 2 10 0.01 1 accuracy binary 0.844 5 ## 6 3 1000 2 5 0.01 0.1 accuracy binary 0.844 5 ## 7 5 1000 2 10 0.01 0.1 accuracy binary 0.843 5 ## 8 3 1000 2 10 0.01 0.1 accuracy binary 0.843 5 ## 9 5 1000 2 5 0.01 0.01 accuracy binary 0.843 5 ## 10 5 1000 2 5 0.01 0.1 accuracy binary 0.843 5 ## # … with 152 more rows, and 2 more variables: std_err \u0026lt;dbl\u0026gt;, .config \u0026lt;chr\u0026gt; The tuning takes some time. There are other ways to explore the parameter space more efficiently. For example, we will use the function dials::grid_max_entropy() in the last section about ensemble modelling. Here, I will use finetune::tune_race_anova.\nlibrary(finetune) xg_tune \u0026lt;- xg_wf %\u0026gt;% tune_race_anova( train_5fold, grid = 50, param_info = xg_model %\u0026gt;% parameters(), metrics = metric_set(accuracy), control = control_race(verbose_elim = TRUE)) Visualize the results.\nautoplot(xg_tune) Collect metrics.\nxg_tune %\u0026gt;% collect_metrics() %\u0026gt;% arrange(desc(mean)) ## # A tibble: 50 × 12 ## mtry trees min_n tree_depth learn_rate loss_reduction .metric .estimator mean n ## \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;int\u0026gt; ## 1 6 856 4 13 1.12e- 2 2.49e- 8 accuracy binary 0.837 5 ## 2 10 1952 6 7 3.36e- 2 2.07e+ 0 accuracy binary 0.829 5 ## 3 3 896 2 5 1.73e- 5 6.97e- 8 accuracy binary 0.826 5 ## 4 14 1122 4 6 1.16e- 6 3.44e+ 0 accuracy binary 0.815 4 ## 5 7 939 8 4 2.88e- 5 1.50e- 4 accuracy binary 0.813 4 ## 6 7 17 10 7 4.54e- 6 6.38e- 3 accuracy binary 0.813 4 ## 7 8 92 9 11 3.60e- 3 3.01e-10 accuracy binary 0.811 4 ## 8 13 1407 15 4 1.48e- 2 9.68e- 4 accuracy binary 0.807 3 ## 9 4 658 11 9 9.97e-10 1.27e- 5 accuracy binary 0.805 3 ## 10 2 628 14 9 1.84e- 6 9.44e- 5 accuracy binary 0.798 3 ## # … with 40 more rows, and 2 more variables: std_err \u0026lt;dbl\u0026gt;, .config \u0026lt;chr\u0026gt; Fit model Use best config to fit model to training data.\nxg_fit \u0026lt;- xg_wf %\u0026gt;% finalize_workflow(select_best(xg_tune)) %\u0026gt;% fit(train) ## [23:39:25] WARNING: amalgamation/../src/learner.cc:1095: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior. Check out accuracy on testing dataset to see if we overfitted.\nxg_fit %\u0026gt;% augment(test, type.predict = \u0026quot;response\u0026quot;) %\u0026gt;% accuracy(survived, .pred_class) ## # A tibble: 1 × 3 ## .metric .estimator .estimate ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; ## 1 accuracy binary 0.812 Check out important features (aka predictors).\nimportances \u0026lt;- xgboost::xgb.importance(model = extract_fit_engine(xg_fit)) importances %\u0026gt;% mutate(Feature = fct_reorder(Feature, Gain)) %\u0026gt;% ggplot(aes(Gain, Feature)) + geom_col() Make predictions Now we’re ready to predict survival for the holdout dataset and submit to Kaggle. Note that I use the whole dataset, not just the training dataset.\nxg_wf %\u0026gt;% finalize_workflow(select_best(xg_tune)) %\u0026gt;% fit(dataset) %\u0026gt;% augment(holdout) %\u0026gt;% select(PassengerId, Survived = .pred_class) %\u0026gt;% mutate(Survived = if_else(Survived == \u0026quot;yes\u0026quot;, 1, 0)) %\u0026gt;% write_csv(\u0026quot;output/titanic/xgboost.csv\u0026quot;) ## [23:39:28] WARNING: amalgamation/../src/learner.cc:1095: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior. I got and accuracy of 0.74162. Cool. Let’s train a random forest model now.\nRandom forests Let’s continue with random forest methods.\nTuning First a recipe.\nrf_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median step_dummy(all_nominal_predictors()) # all factors var are split into binary terms (factor disj coding) Then specify a random forest model.\nrf_model \u0026lt;- rand_forest(mode = \u0026quot;classification\u0026quot;, # binary response engine = \u0026quot;ranger\u0026quot;, # by default mtry = tune(), trees = tune(), min_n = tune()) # parameters to be tuned Now set our workflow.\nrf_wf \u0026lt;- workflow() %\u0026gt;% add_model(rf_model) %\u0026gt;% add_recipe(rf_rec) Use cross-validation to evaluate our model with different param config.\nrf_tune \u0026lt;- rf_wf %\u0026gt;% tune_race_anova( train_5fold, grid = 50, param_info = rf_model %\u0026gt;% parameters(), metrics = metric_set(accuracy), control = control_race(verbose_elim = TRUE)) Visualize the results.\nautoplot(rf_tune) Collect metrics.\nrf_tune %\u0026gt;% collect_metrics() %\u0026gt;% arrange(desc(mean)) ## # A tibble: 50 × 9 ## mtry trees min_n .metric .estimator mean n std_err .config ## \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; ## 1 14 1554 5 accuracy binary 0.837 5 0.00656 Preprocessor1_Model49 ## 2 4 774 18 accuracy binary 0.837 5 0.0133 Preprocessor1_Model50 ## 3 5 1736 8 accuracy binary 0.834 5 0.0111 Preprocessor1_Model46 ## 4 8 1322 5 accuracy binary 0.832 5 0.00713 Preprocessor1_Model41 ## 5 2 1078 30 accuracy binary 0.831 5 0.00727 Preprocessor1_Model12 ## 6 4 1892 14 accuracy binary 0.831 5 0.00886 Preprocessor1_Model39 ## 7 8 962 7 accuracy binary 0.829 5 0.00742 Preprocessor1_Model43 ## 8 7 946 4 accuracy binary 0.826 5 0.00452 Preprocessor1_Model23 ## 9 7 1262 3 accuracy binary 0.826 5 0.00710 Preprocessor1_Model47 ## 10 9 544 9 accuracy binary 0.825 5 0.00673 Preprocessor1_Model11 ## # … with 40 more rows Fit model Use best config to fit model to training data.\nrf_fit \u0026lt;- rf_wf %\u0026gt;% finalize_workflow(select_best(rf_tune)) %\u0026gt;% fit(train) Check out accuracy on testing dataset to see if we overfitted.\nrf_fit %\u0026gt;% augment(test, type.predict = \u0026quot;response\u0026quot;) %\u0026gt;% accuracy(survived, .pred_class) ## # A tibble: 1 × 3 ## .metric .estimator .estimate ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; ## 1 accuracy binary 0.786 Check out important features (aka predictors).\nlibrary(vip) finalize_model( x = rf_model, parameters = select_best(rf_tune)) %\u0026gt;% set_engine(\u0026quot;ranger\u0026quot;, importance = \u0026quot;permutation\u0026quot;) %\u0026gt;% fit(survived ~ ., data = juice(prep(rf_rec))) %\u0026gt;% vip(geom = \u0026quot;point\u0026quot;) Make predictions Now we’re ready to predict survival for the holdout dataset and submit to Kaggle.\nrf_wf %\u0026gt;% finalize_workflow(select_best(rf_tune)) %\u0026gt;% fit(dataset) %\u0026gt;% augment(holdout) %\u0026gt;% select(PassengerId, Survived = .pred_class) %\u0026gt;% mutate(Survived = if_else(Survived == \u0026quot;yes\u0026quot;, 1, 0)) %\u0026gt;% write_csv(\u0026quot;output/titanic/randomforest.csv\u0026quot;) I got and accuracy of 0.77990, a bit better than gradient boosting.\nLet’s continue with cat boosting methods.\nGradient boosting algorithms - catboost Tuning Set up defaults.\nmset \u0026lt;- metric_set(accuracy) # metric is accuracy control \u0026lt;- control_grid(save_workflow = TRUE, save_pred = TRUE, extract = extract_model) # grid for tuning First a recipe.\ncb_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median step_dummy(all_nominal_predictors()) # all factors var are split into binary terms (factor disj coding) Then specify a cat boosting model.\nlibrary(treesnip) cb_model \u0026lt;- boost_tree(mode = \u0026quot;classification\u0026quot;, engine = \u0026quot;catboost\u0026quot;, mtry = tune(), trees = tune(), min_n = tune(), tree_depth = tune(), learn_rate = tune()) # parameters to be tuned Now set our workflow.\ncb_wf \u0026lt;- workflow() %\u0026gt;% add_model(cb_model) %\u0026gt;% add_recipe(cb_rec) Use cross-validation to evaluate our model with different param config.\ncb_tune \u0026lt;- cb_wf %\u0026gt;% tune_race_anova( train_5fold, grid = 30, param_info = cb_model %\u0026gt;% parameters(), metrics = metric_set(accuracy), control = control_race(verbose_elim = TRUE)) Visualize the results.\nautoplot(cb_tune) Collect metrics.\ncb_tune %\u0026gt;% collect_metrics() %\u0026gt;% arrange(desc(mean)) ## # A tibble: 30 × 11 ## mtry trees min_n tree_depth learn_rate .metric .estimator mean n std_err .config ## \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; ## 1 1 1787 25 7 6.84e- 3 accuracy binary 0.835 5 0.0125 Preproc… ## 2 12 1885 18 15 2.06e- 3 accuracy binary 0.831 5 0.00602 Preproc… ## 3 13 1278 5 6 2.10e- 2 accuracy binary 0.826 5 0.00431 Preproc… ## 4 3 1681 22 5 5.42e- 3 accuracy binary 0.825 5 0.00507 Preproc… ## 5 9 303 2 8 9.94e- 2 accuracy binary 0.820 4 0.0120 Preproc… ## 6 11 1201 24 12 3.77e- 2 accuracy binary 0.812 3 0.00868 Preproc… ## 7 11 634 35 11 1.30e- 3 accuracy binary 0.805 3 0.0242 Preproc… ## 8 7 1648 4 7 2.38e- 4 accuracy binary 0.737 3 0.0115 Preproc… ## 9 1 1427 27 10 1.74e- 6 accuracy binary 0.378 3 0.0152 Preproc… ## 10 2 940 19 3 3.20e-10 accuracy binary 0.378 3 0.0152 Preproc… ## # … with 20 more rows Fit model Use best config to fit model to training data.\ncb_fit \u0026lt;- cb_wf %\u0026gt;% finalize_workflow(select_best(cb_tune)) %\u0026gt;% fit(train) Check out accuracy on testing dataset to see if we overfitted.\ncb_fit %\u0026gt;% augment(test, type.predict = \u0026quot;response\u0026quot;) %\u0026gt;% accuracy(survived, .pred_class) ## # A tibble: 1 × 3 ## .metric .estimator .estimate ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; ## 1 accuracy binary 0.808 Make predictions Now we’re ready to predict survival for the holdout dataset and submit to Kaggle.\ncb_wf %\u0026gt;% finalize_workflow(select_best(cb_tune)) %\u0026gt;% fit(dataset) %\u0026gt;% augment(holdout) %\u0026gt;% select(PassengerId, Survived = .pred_class) %\u0026gt;% mutate(Survived = if_else(Survived == \u0026quot;yes\u0026quot;, 1, 0)) %\u0026gt;% write_csv(\u0026quot;output/titanic/catboost.csv\u0026quot;) I got and accuracy of 0.76076. Cool.\nRegularization methods Let’s continue with elastic net regularization.\nTuning First a recipe.\nen_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median step_normalize(all_numeric_predictors()) %\u0026gt;% # normalize step_dummy(all_nominal_predictors()) # all factors var are split into binary terms (factor disj coding) Then specify a regularization model. We tune parameter mixture, with ridge regression for mixture = 0, and lasso for mixture = 1.\nen_model \u0026lt;- logistic_reg(penalty = tune(), mixture = tune()) %\u0026gt;% # param to be tuned set_engine(\u0026quot;glmnet\u0026quot;) %\u0026gt;% # elastic net set_mode(\u0026quot;classification\u0026quot;) # binary response Now set our workflow.\nen_wf \u0026lt;- workflow() %\u0026gt;% add_model(en_model) %\u0026gt;% add_recipe(en_rec) Use cross-validation to evaluate our model with different param config.\nen_tune \u0026lt;- en_wf %\u0026gt;% tune_grid(train_5fold, metrics = mset, control = control, grid = crossing(penalty = 10 ^ seq(-8, -.5, .5), mixture = seq(0, 1, length.out = 10))) Visualize the results.\nautoplot(en_tune) Collect metrics.\nen_tune %\u0026gt;% collect_metrics() %\u0026gt;% arrange(desc(mean)) ## # A tibble: 160 × 8 ## penalty mixture .metric .estimator mean n std_err .config ## \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; ## 1 0.00000001 0.111 accuracy binary 0.831 5 0.0112 Preprocessor1_Model017 ## 2 0.0000000316 0.111 accuracy binary 0.831 5 0.0112 Preprocessor1_Model018 ## 3 0.0000001 0.111 accuracy binary 0.831 5 0.0112 Preprocessor1_Model019 ## 4 0.000000316 0.111 accuracy binary 0.831 5 0.0112 Preprocessor1_Model020 ## 5 0.000001 0.111 accuracy binary 0.831 5 0.0112 Preprocessor1_Model021 ## 6 0.00000316 0.111 accuracy binary 0.831 5 0.0112 Preprocessor1_Model022 ## 7 0.00001 0.111 accuracy binary 0.831 5 0.0112 Preprocessor1_Model023 ## 8 0.0000316 0.111 accuracy binary 0.831 5 0.0112 Preprocessor1_Model024 ## 9 0.0001 0.111 accuracy binary 0.831 5 0.0112 Preprocessor1_Model025 ## 10 0.000316 0.111 accuracy binary 0.831 5 0.0112 Preprocessor1_Model026 ## # … with 150 more rows Fit model Use best config to fit model to training data.\nen_fit \u0026lt;- en_wf %\u0026gt;% finalize_workflow(select_best(en_tune)) %\u0026gt;% fit(train) Check out accuracy on testing dataset to see if we overfitted.\nen_fit %\u0026gt;% augment(test, type.predict = \u0026quot;response\u0026quot;) %\u0026gt;% accuracy(survived, .pred_class) ## # A tibble: 1 × 3 ## .metric .estimator .estimate ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; ## 1 accuracy binary 0.826 Check out important features (aka predictors).\nlibrary(broom) en_fit$fit$fit$fit %\u0026gt;% tidy() %\u0026gt;% filter(lambda \u0026gt;= select_best(en_tune)$penalty) %\u0026gt;% filter(lambda == min(lambda), term != \u0026quot;(Intercept)\u0026quot;) %\u0026gt;% mutate(term = fct_reorder(term, estimate)) %\u0026gt;% ggplot(aes(estimate, term, fill = estimate \u0026gt; 0)) + geom_col() + theme(legend.position = \u0026quot;none\u0026quot;) Make predictions Now we’re ready to predict survival for the holdout dataset and submit to Kaggle.\nen_wf %\u0026gt;% finalize_workflow(select_best(en_tune)) %\u0026gt;% fit(dataset) %\u0026gt;% augment(holdout) %\u0026gt;% select(PassengerId, Survived = .pred_class) %\u0026gt;% mutate(Survived = if_else(Survived == \u0026quot;yes\u0026quot;, 1, 0)) %\u0026gt;% write_csv(\u0026quot;output/titanic/regularization.csv\u0026quot;) I got and accuracy of 0.76315.\nLogistic regression And what about a good old-fashioned logistic regression (not a ML algo)?\nFirst a recipe.\nlogistic_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median step_normalize(all_numeric_predictors()) %\u0026gt;% # normalize step_dummy(all_nominal_predictors()) # all factors var are split into binary terms (factor disj coding) Then specify a logistic regression.\nlogistic_model \u0026lt;- logistic_reg() %\u0026gt;% # no param to be tuned set_engine(\u0026quot;glm\u0026quot;) %\u0026gt;% # elastic net set_mode(\u0026quot;classification\u0026quot;) # binary response Now set our workflow.\nlogistic_wf \u0026lt;- workflow() %\u0026gt;% add_model(logistic_model) %\u0026gt;% add_recipe(logistic_rec) Fit model.\nlogistic_fit \u0026lt;- logistic_wf %\u0026gt;% fit(train) Inspect significant features (aka predictors).\ntidy(logistic_fit, exponentiate = TRUE) %\u0026gt;% filter(p.value \u0026lt; 0.05) ## # A tibble: 6 × 5 ## term estimate std.error statistic p.value ## \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; ## 1 age 1.35 0.152 1.98 0.0473 ## 2 familyage 2.97 0.223 4.88 0.00000105 ## 3 class_first 0.136 0.367 -5.43 0.0000000559 ## 4 class_second 0.386 0.298 -3.19 0.00145 ## 5 title_Mr. 51.0 0.684 5.75 0.00000000912 ## 6 title_Other 54.7 0.991 4.04 0.0000538 Same thing, but graphically.\nlibrary(broom) logistic_fit %\u0026gt;% tidy() %\u0026gt;% mutate(term = fct_reorder(term, estimate)) %\u0026gt;% ggplot(aes(estimate, term, fill = estimate \u0026gt; 0)) + geom_col() + theme(legend.position = \u0026quot;none\u0026quot;) Check out accuracy on testing dataset to see if we overfitted.\nlogistic_fit %\u0026gt;% augment(test, type.predict = \u0026quot;response\u0026quot;) %\u0026gt;% accuracy(survived, .pred_class) ## # A tibble: 1 × 3 ## .metric .estimator .estimate ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; ## 1 accuracy binary 0.821 Confusion matrix.\nlogistic_fit %\u0026gt;% augment(test, type.predict = \u0026quot;response\u0026quot;) %\u0026gt;% conf_mat(survived, .pred_class) ## Truth ## Prediction yes no ## yes 59 13 ## no 27 125 ROC curve.\nlogistic_fit %\u0026gt;% augment(test, type.predict = \u0026quot;response\u0026quot;) %\u0026gt;% roc_curve(truth = survived, estimate = .pred_yes) %\u0026gt;% autoplot() Now we’re ready to predict survival for the holdout dataset and submit to Kaggle.\nlogistic_wf %\u0026gt;% fit(dataset) %\u0026gt;% augment(holdout) %\u0026gt;% select(PassengerId, Survived = .pred_class) %\u0026gt;% mutate(Survived = if_else(Survived == \u0026quot;yes\u0026quot;, 1, 0)) %\u0026gt;% write_csv(\u0026quot;output/titanic/logistic.csv\u0026quot;) I got and accuracy of 0.76076. Oldies but goodies!\nNeural networks We go on with neural networks.\nTuning Set up defaults.\nmset \u0026lt;- metric_set(accuracy) # metric is accuracy control \u0026lt;- control_grid(save_workflow = TRUE, save_pred = TRUE, extract = extract_model) # grid for tuning First a recipe.\nnn_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median step_normalize(all_numeric_predictors()) %\u0026gt;% step_dummy(all_nominal_predictors()) # all factors var are split into binary terms (factor disj coding) Then specify a neural network.\nnn_model \u0026lt;- mlp(epochs = tune(), hidden_units = tune(), dropout = tune()) %\u0026gt;% # param to be tuned set_mode(\u0026quot;classification\u0026quot;) %\u0026gt;% # binary response var set_engine(\u0026quot;keras\u0026quot;, verbose = 0) Now set our workflow.\nnn_wf \u0026lt;- workflow() %\u0026gt;% add_model(nn_model) %\u0026gt;% add_recipe(nn_rec) Use cross-validation to evaluate our model with different param config.\nnn_tune \u0026lt;- nn_wf %\u0026gt;% tune_race_anova( train_5fold, grid = 30, param_info = nn_model %\u0026gt;% parameters(), metrics = metric_set(accuracy), control = control_race(verbose_elim = TRUE)) Visualize the results.\nautoplot(nn_tune) Collect metrics.\nnn_tune %\u0026gt;% collect_metrics() %\u0026gt;% arrange(desc(mean)) ## # A tibble: 30 × 9 ## hidden_units dropout epochs .metric .estimator mean n std_err .config ## \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;int\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; ## 1 10 0.484 405 accuracy binary 0.838 5 0.0150 Preprocessor1_Model… ## 2 7 0.0597 220 accuracy binary 0.834 5 0.00881 Preprocessor1_Model… ## 3 4 0.536 629 accuracy binary 0.834 5 0.0154 Preprocessor1_Model… ## 4 9 0.198 768 accuracy binary 0.832 5 0.0146 Preprocessor1_Model… ## 5 6 0.752 822 accuracy binary 0.832 5 0.0112 Preprocessor1_Model… ## 6 9 0.406 293 accuracy binary 0.831 5 0.0145 Preprocessor1_Model… ## 7 4 0.00445 871 accuracy binary 0.831 5 0.0150 Preprocessor1_Model… ## 8 4 0.293 353 accuracy binary 0.831 5 0.0117 Preprocessor1_Model… ## 9 10 0.675 935 accuracy binary 0.831 5 0.0141 Preprocessor1_Model… ## 10 7 0.128 580 accuracy binary 0.831 5 0.0151 Preprocessor1_Model… ## # … with 20 more rows Fit model Use best config to fit model to training data.\nnn_fit \u0026lt;- nn_wf %\u0026gt;% finalize_workflow(select_best(nn_tune)) %\u0026gt;% fit(train) Check out accuracy on testing dataset to see if we overfitted.\nnn_fit %\u0026gt;% augment(test, type.predict = \u0026quot;response\u0026quot;) %\u0026gt;% accuracy(survived, .pred_class) ## # A tibble: 1 × 3 ## .metric .estimator .estimate ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; ## 1 accuracy binary 0.817 Make predictions Now we’re ready to predict survival for the holdout dataset and submit to Kaggle.\nnn_wf %\u0026gt;% finalize_workflow(select_best(nn_tune)) %\u0026gt;% fit(train) %\u0026gt;% augment(holdout) %\u0026gt;% select(PassengerId, Survived = .pred_class) %\u0026gt;% mutate(Survived = if_else(Survived == \u0026quot;yes\u0026quot;, 1, 0)) %\u0026gt;% write_csv(\u0026quot;output/titanic/nn.csv\u0026quot;) I got and accuracy of 0.78708. My best score so far.\nSupport vector machines We go on with support vector machines.\nTuning Set up defaults.\nmset \u0026lt;- metric_set(accuracy) # metric is accuracy control \u0026lt;- control_grid(save_workflow = TRUE, save_pred = TRUE, extract = extract_model) # grid for tuning First a recipe.\nsvm_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median # remove any zero variance predictors step_zv(all_predictors()) %\u0026gt;% # remove any linear combinations step_lincomb(all_numeric()) %\u0026gt;% step_dummy(all_nominal_predictors()) # all factors var are split into binary terms (factor disj coding) Then specify a svm.\nsvm_model \u0026lt;- svm_rbf(cost = tune(), rbf_sigma = tune()) %\u0026gt;% # param to be tuned set_mode(\u0026quot;classification\u0026quot;) %\u0026gt;% # binary response var set_engine(\u0026quot;kernlab\u0026quot;) Now set our workflow.\nsvm_wf \u0026lt;- workflow() %\u0026gt;% add_model(svm_model) %\u0026gt;% add_recipe(svm_rec) Use cross-validation to evaluate our model with different param config.\nsvm_tune \u0026lt;- svm_wf %\u0026gt;% tune_race_anova( train_5fold, grid = 30, param_info = svm_model %\u0026gt;% parameters(), metrics = metric_set(accuracy), control = control_race(verbose_elim = TRUE)) Visualize the results.\nautoplot(svm_tune) Collect metrics.\nsvm_tune %\u0026gt;% collect_metrics() %\u0026gt;% arrange(desc(mean)) ## # A tibble: 30 × 8 ## cost rbf_sigma .metric .estimator mean n std_err .config ## \u0026lt;dbl\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; ## 1 17.7 0.00183 accuracy binary 0.829 5 0.0101 Preprocessor1_Model05 ## 2 0.741 0.0506 accuracy binary 0.823 5 0.0129 Preprocessor1_Model14 ## 3 1.77 0.0429 accuracy binary 0.820 5 0.0143 Preprocessor1_Model08 ## 4 0.229 0.0197 accuracy binary 0.808 3 0.00626 Preprocessor1_Model23 ## 5 1.15 0.00285 accuracy binary 0.795 3 0.00867 Preprocessor1_Model28 ## 6 0.00182 0.00000203 accuracy binary 0.613 3 0.0171 Preprocessor1_Model01 ## 7 0.0477 0.714 accuracy binary 0.613 3 0.0171 Preprocessor1_Model02 ## 8 6.60 0.0000294 accuracy binary 0.613 3 0.0171 Preprocessor1_Model03 ## 9 0.00254 0.000636 accuracy binary 0.613 3 0.0171 Preprocessor1_Model04 ## 10 0.00544 0.0000000647 accuracy binary 0.613 3 0.0171 Preprocessor1_Model06 ## # … with 20 more rows Fit model Use best config to fit model to training data.\nsvm_fit \u0026lt;- svm_wf %\u0026gt;% finalize_workflow(select_best(svm_tune)) %\u0026gt;% fit(train) Check out accuracy on testing dataset to see if we overfitted.\nsvm_fit %\u0026gt;% augment(test, type.predict = \u0026quot;response\u0026quot;) %\u0026gt;% accuracy(survived, .pred_class) ## # A tibble: 1 × 3 ## .metric .estimator .estimate ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; ## 1 accuracy binary 0.826 Make predictions Now we’re ready to predict survival for the holdout dataset and submit to Kaggle.\nsvm_wf %\u0026gt;% finalize_workflow(select_best(svm_tune)) %\u0026gt;% fit(dataset) %\u0026gt;% augment(holdout) %\u0026gt;% select(PassengerId, Survived = .pred_class) %\u0026gt;% mutate(Survived = if_else(Survived == \u0026quot;yes\u0026quot;, 1, 0)) %\u0026gt;% write_csv(\u0026quot;output/titanic/svm.csv\u0026quot;) I got and accuracy of 0.77511.\nDecision trees We go on with decision trees.\nTuning Set up defaults.\nmset \u0026lt;- metric_set(accuracy) # metric is accuracy control \u0026lt;- control_grid(save_workflow = TRUE, save_pred = TRUE, extract = extract_model) # grid for tuning First a recipe.\ndt_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median step_zv(all_predictors()) %\u0026gt;% step_dummy(all_nominal_predictors()) # all factors var are split into binary terms (factor disj coding) Then specify a decision tree model.\nlibrary(baguette) dt_model \u0026lt;- bag_tree(cost_complexity = tune(), tree_depth = tune(), min_n = tune()) %\u0026gt;% # param to be tuned set_engine(\u0026quot;rpart\u0026quot;, times = 25) %\u0026gt;% # nb bootstraps set_mode(\u0026quot;classification\u0026quot;) # binary response var Now set our workflow.\ndt_wf \u0026lt;- workflow() %\u0026gt;% add_model(dt_model) %\u0026gt;% add_recipe(dt_rec) Use cross-validation to evaluate our model with different param config.\ndt_tune \u0026lt;- dt_wf %\u0026gt;% tune_race_anova( train_5fold, grid = 30, param_info = dt_model %\u0026gt;% parameters(), metrics = metric_set(accuracy), control = control_race(verbose_elim = TRUE)) Visualize the results.\nautoplot(dt_tune) Collect metrics.\ndt_tune %\u0026gt;% collect_metrics() %\u0026gt;% arrange(desc(mean)) ## # A tibble: 30 × 9 ## cost_complexity tree_depth min_n .metric .estimator mean n std_err .config ## \u0026lt;dbl\u0026gt; \u0026lt;int\u0026gt; \u0026lt;int\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;int\u0026gt; \u0026lt;dbl\u0026gt; \u0026lt;chr\u0026gt; ## 1 3.02e- 7 13 7 accuracy binary 0.832 5 0.0130 Preprocessor1_… ## 2 3.43e- 3 14 4 accuracy binary 0.829 5 0.0145 Preprocessor1_… ## 3 5.75e- 5 5 7 accuracy binary 0.828 5 0.0126 Preprocessor1_… ## 4 2.13e- 6 14 5 accuracy binary 0.828 5 0.0108 Preprocessor1_… ## 5 1.34e- 5 5 35 accuracy binary 0.823 5 0.00654 Preprocessor1_… ## 6 4.47e- 5 13 10 accuracy binary 0.823 5 0.0120 Preprocessor1_… ## 7 1.42e- 2 4 25 accuracy binary 0.822 5 0.0121 Preprocessor1_… ## 8 4.54e-10 10 36 accuracy binary 0.822 5 0.0147 Preprocessor1_… ## 9 8.10e- 8 11 32 accuracy binary 0.822 5 0.0143 Preprocessor1_… ## 10 3.43e- 4 11 21 accuracy binary 0.820 5 0.0207 Preprocessor1_… ## # … with 20 more rows Fit model Use best config to fit model to training data.\ndt_fit \u0026lt;- dt_wf %\u0026gt;% finalize_workflow(select_best(dt_tune)) %\u0026gt;% fit(train) Check out accuracy on testing dataset to see if we overfitted.\ndt_fit %\u0026gt;% augment(test, type.predict = \u0026quot;response\u0026quot;) %\u0026gt;% accuracy(survived, .pred_class) ## # A tibble: 1 × 3 ## .metric .estimator .estimate ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; ## 1 accuracy binary 0.808 Make predictions Now we’re ready to predict survival for the holdout dataset and submit to Kaggle.\ndt_wf %\u0026gt;% finalize_workflow(select_best(dt_tune)) %\u0026gt;% fit(dataset) %\u0026gt;% augment(holdout) %\u0026gt;% select(PassengerId, Survived = .pred_class) %\u0026gt;% mutate(Survived = if_else(Survived == \u0026quot;yes\u0026quot;, 1, 0)) %\u0026gt;% write_csv(\u0026quot;output/titanic/dt.csv\u0026quot;) I got and accuracy of 0.76794.\nStacked ensemble modelling Let’s do some ensemble modelling with all algo but logistic and catboost. Tune again with a probability-based metric. Start with xgboost.\nlibrary(finetune) library(stacks) # xgboost xg_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median step_dummy(all_nominal_predictors()) # all factors var are split into binary terms (factor disj coding) xg_model \u0026lt;- boost_tree(mode = \u0026quot;classification\u0026quot;, # binary response trees = tune(), mtry = tune(), tree_depth = tune(), learn_rate = tune(), loss_reduction = tune(), min_n = tune()) # parameters to be tuned xg_wf \u0026lt;- workflow() %\u0026gt;% add_model(xg_model) %\u0026gt;% add_recipe(xg_rec) xg_grid \u0026lt;- grid_latin_hypercube( trees(), finalize(mtry(), train), tree_depth(), learn_rate(), loss_reduction(), min_n(), size = 30) xg_tune \u0026lt;- xg_wf %\u0026gt;% tune_grid( resamples = train_5fold, grid = xg_grid, metrics = metric_set(roc_auc), control = control_stack_grid()) Then random forests.\n# random forest rf_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% step_dummy(all_nominal_predictors()) rf_model \u0026lt;- rand_forest(mode = \u0026quot;classification\u0026quot;, # binary response engine = \u0026quot;ranger\u0026quot;, # by default mtry = tune(), trees = tune(), min_n = tune()) # parameters to be tuned rf_wf \u0026lt;- workflow() %\u0026gt;% add_model(rf_model) %\u0026gt;% add_recipe(rf_rec) rf_grid \u0026lt;- grid_latin_hypercube( finalize(mtry(), train), trees(), min_n(), size = 30) rf_tune \u0026lt;- rf_wf %\u0026gt;% tune_grid( resamples = train_5fold, grid = rf_grid, metrics = metric_set(roc_auc), control = control_stack_grid()) Regularisation methods (between ridge and lasso).\n# regularization methods en_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median step_normalize(all_numeric_predictors()) %\u0026gt;% # normalize step_dummy(all_nominal_predictors()) en_model \u0026lt;- logistic_reg(penalty = tune(), mixture = tune()) %\u0026gt;% # param to be tuned set_engine(\u0026quot;glmnet\u0026quot;) %\u0026gt;% # elastic net set_mode(\u0026quot;classification\u0026quot;) # binary response en_wf \u0026lt;- workflow() %\u0026gt;% add_model(en_model) %\u0026gt;% add_recipe(en_rec) en_grid \u0026lt;- grid_latin_hypercube( penalty(), mixture(), size = 30) en_tune \u0026lt;- en_wf %\u0026gt;% tune_grid( resamples = train_5fold, grid = en_grid, metrics = metric_set(roc_auc), control = control_stack_grid()) Neural networks (takes time, so pick only a few values for illustration purpose).\n# neural networks nn_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% # replace missing value by median step_normalize(all_numeric_predictors()) %\u0026gt;% step_dummy(all_nominal_predictors()) nn_model \u0026lt;- mlp(epochs = tune(), hidden_units = 2, dropout = tune()) %\u0026gt;% # param to be tuned set_mode(\u0026quot;classification\u0026quot;) %\u0026gt;% # binary response var set_engine(\u0026quot;keras\u0026quot;, verbose = 0) nn_wf \u0026lt;- workflow() %\u0026gt;% add_model(nn_model) %\u0026gt;% add_recipe(nn_rec) # nn_grid \u0026lt;- grid_latin_hypercube( # epochs(), # hidden_units(), # dropout(), # size = 10) nn_tune \u0026lt;- nn_wf %\u0026gt;% tune_grid( resamples = train_5fold, grid = crossing(dropout = c(0.1, 0.2), epochs = c(250, 500, 1000)), # nn_grid metrics = metric_set(roc_auc), control = control_stack_grid()) #autoplot(nn_tune) Support vector machines.\n# support vector machines svm_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% step_normalize(all_numeric_predictors()) %\u0026gt;% step_dummy(all_nominal_predictors()) svm_model \u0026lt;- svm_rbf(cost = tune(), rbf_sigma = tune()) %\u0026gt;% # param to be tuned set_mode(\u0026quot;classification\u0026quot;) %\u0026gt;% # binary response var set_engine(\u0026quot;kernlab\u0026quot;) svm_wf \u0026lt;- workflow() %\u0026gt;% add_model(svm_model) %\u0026gt;% add_recipe(svm_rec) svm_grid \u0026lt;- grid_latin_hypercube( cost(), rbf_sigma(), size = 30) svm_tune \u0026lt;- svm_wf %\u0026gt;% tune_grid( resamples = train_5fold, grid = svm_grid, metrics = metric_set(roc_auc), control = control_stack_grid()) Last, decision trees.\n# decision trees dt_rec \u0026lt;- recipe(survived ~ ., data = train) %\u0026gt;% step_impute_median(all_numeric()) %\u0026gt;% step_zv(all_predictors()) %\u0026gt;% step_dummy(all_nominal_predictors()) library(baguette) dt_model \u0026lt;- bag_tree(cost_complexity = tune(), tree_depth = tune(), min_n = tune()) %\u0026gt;% # param to be tuned set_engine(\u0026quot;rpart\u0026quot;, times = 25) %\u0026gt;% # nb bootstraps set_mode(\u0026quot;classification\u0026quot;) # binary response var dt_wf \u0026lt;- workflow() %\u0026gt;% add_model(dt_model) %\u0026gt;% add_recipe(dt_rec) dt_grid \u0026lt;- grid_latin_hypercube( cost_complexity(), tree_depth(), min_n(), size = 30) dt_tune \u0026lt;- dt_wf %\u0026gt;% tune_grid( resamples = train_5fold, grid = dt_grid, metrics = metric_set(roc_auc), control = control_stack_grid()) Get best config.\nxg_best \u0026lt;- xg_tune %\u0026gt;% filter_parameters(parameters = select_best(xg_tune)) rf_best \u0026lt;- rf_tune %\u0026gt;% filter_parameters(parameters = select_best(rf_tune)) en_best \u0026lt;- en_tune %\u0026gt;% filter_parameters(parameters = select_best(en_tune)) nn_best \u0026lt;- nn_tune %\u0026gt;% filter_parameters(parameters = select_best(nn_tune)) svm_best \u0026lt;- svm_tune %\u0026gt;% filter_parameters(parameters = select_best(svm_tune)) dt_best \u0026lt;- dt_tune %\u0026gt;% filter_parameters(parameters = select_best(dt_tune)) Do the stacked ensemble modelling.\nPile all models together.\nblended \u0026lt;- stacks() %\u0026gt;% # initialize add_candidates(en_best) %\u0026gt;% # add regularization model add_candidates(xg_best) %\u0026gt;% # add gradient boosting add_candidates(rf_best) %\u0026gt;% # add random forest add_candidates(nn_best) %\u0026gt;% # add neural network add_candidates(svm_best) %\u0026gt;% # add svm add_candidates(dt_best) # add decision trees blended ## # A data stack with 6 model definitions and 6 candidate members: ## # en_best: 1 model configuration ## # xg_best: 1 model configuration ## # rf_best: 1 model configuration ## # nn_best: 1 model configuration ## # svm_best: 1 model configuration ## # dt_best: 1 model configuration ## # Outcome: survived (factor) Fit regularized model.\nblended_fit \u0026lt;- blended %\u0026gt;% blend_predictions() # fit regularized model Visualise penalized model. Note that neural networks are dropped, despite achieving best score when used in isolation. I’ll have to dig into that.\nautoplot(blended_fit) autoplot(blended_fit, type = \u0026quot;members\u0026quot;) autoplot(blended_fit, type = \u0026quot;weights\u0026quot;) Fit candidate members with non-zero stacking coef with full training dataset.\nblended_regularized \u0026lt;- blended_fit %\u0026gt;% fit_members() blended_regularized ## # A tibble: 3 × 3 ## member type weight ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; ## 1 .pred_no_en_best_1_12 logistic_reg 2.67 ## 2 .pred_no_dt_best_1_11 bag_tree 1.95 ## 3 .pred_no_rf_best_1_05 rand_forest 0.922 Perf on testing dataset?\ntest %\u0026gt;% bind_cols(predict(blended_regularized, .)) %\u0026gt;% accuracy(survived, .pred_class) ## # A tibble: 1 × 3 ## .metric .estimator .estimate ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;dbl\u0026gt; ## 1 accuracy binary 0.826 Now predict.\nholdout %\u0026gt;% bind_cols(predict(blended_regularized, .)) %\u0026gt;% select(PassengerId, Survived = .pred_class) %\u0026gt;% mutate(Survived = if_else(Survived == \u0026quot;yes\u0026quot;, 1, 0)) %\u0026gt;% write_csv(\u0026quot;output/titanic/stacked.csv\u0026quot;) I got an 0.76076 accuracy.\nConclusions I covered several ML algorithms and logistic regression with the awesome tidymodels metapackage in R. My scores at predicting Titanic survivors were ok I guess. Some folks on Kaggle got a perfect accuracy, so there is always room for improvement. Maybe better tuning, better features (or predictors) or other algorithms would increase accuracy. Of course, I forgot to use set.seed() so results are not exactly reproducible.\nThis post was also published on https://www.r-bloggers.com.\u0026#160;\u0026#x21a9;\u0026#xfe0e;\n","date":1628812800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1628812800,"objectID":"6d1999265d6ae3cfa97cff110ffd9567","permalink":"https://oliviergimenez.github.io/blog/learning-machine-learning/","publishdate":"2021-08-13T00:00:00Z","relpermalink":"/blog/learning-machine-learning/","section":"blog","summary":"I would like to familiarize myself with machine learning (ML) techniques in R. So I have been reading and learning by doing. I thought I\u0026rsquo;d share my experience for others who\u0026rsquo;d like to give it a try1.\n","tags":["machine learning","tidymodels","R","rstats"],"title":"Experimenting with machine learning in R with tidymodels and the Kaggle titanic dataset","type":"blog"},{"authors":null,"categories":null,"content":"I use R to retrieve some data from Twitter, do some exploratory data analysis and visualisation and examine a network of followers.\nMotivation I use Twitter to get live updates of what my follow scientists are up to, to communicate about my students\u0026rsquo; awesome work and to share material that I hope is useful to some people1.\nRecently, I reached 5,000 followers and I thought I\u0026rsquo;d spend some time trying to know better who they/you are. To do so, I use R to retrieve some data from Twitter using rtweet, do some data exploration and visualisation using the tidyverse and examine my network of followers with tidygraph, ggraph and igraph. Data and codes are available from https://github.com/oliviergimenez/sna-twitter.\nTo reproduce the analyses below, you will need to access a Twitter API (application programming interface) to retrieve the information about your followers. In brief, an API is an intermediary application that allows applications to talk to each other. To access the Twitter APIs, you need a developer account for which you may apply at https://developer.twitter.com/en. There is a short form to fill in, and it takes less than a day to get an answer.\nBelow I rely heavily on the code shared by Joe Cristian through the Algoritma Technical Blog at https://algotech.netlify.app/blog/social-network-analysis-in-r/. Kuddos and credits to him.\nData retrieving We load the rtweet package to work with Twitter from R.\n#devtools::install_github(\u0026quot;ropensci/rtweet\u0026quot;) library(rtweet) We also load the tidyverse for data manipulation and visualisation.\nlibrary(tidyverse) theme_set(theme_light(base_size = 14)) First we need to get credentials. The usual rtweet sequence with rtweet_app, auth_save and auth_as is supposed to work (see here), but the Twitter API kept failing (error 401) for me. Tried a few things, in vain. I will use the deprecated create_token function instead. You might need to change the defaults of your Twitter app from \u0026ldquo;Read only\u0026rdquo; to \u0026ldquo;Read, write and access direct messages\u0026rdquo;.\nEnter my API keys and access tokens (not shown), then authenticate.\ntoken \u0026lt;- create_token( app = \u0026quot;sna-twitter-network-5k\u0026quot;, consumer_key = api_key, consumer_secret = api_secret_key, access_token = access_token, access_secret = access_token_secret) Get my Twitter info with my description, number of followers, number of likes, etc.\nog \u0026lt;- lookup_users(\u0026quot;oaggimenez\u0026quot;) str(og, max = 2) ## 'data.frame':\t1 obs. of 21 variables: ## $ id : num 7.51e+17 ## $ id_str : chr \u0026quot;750892662224453632\u0026quot; ## $ name : chr \u0026quot;Olivier Gimenez \\U0001f596\u0026quot; ## $ screen_name : chr \u0026quot;oaggimenez\u0026quot; ## $ location : chr \u0026quot;Montpellier, France\u0026quot; ## $ description : chr \u0026quot;Scientist (he/him) @CNRS @cefemontpellier @twitthair1 • Grown statistician • Improvised ecologist • Sociologist\u0026quot;| __truncated__ ## $ url : chr \u0026quot;https://t.co/l7NImYeGdY\u0026quot; ## $ protected : logi FALSE ## $ followers_count : int 5042 ## $ friends_count : int 2191 ## $ listed_count : int 68 ## $ created_at : chr \u0026quot;Thu Jul 07 03:22:16 +0000 2016\u0026quot; ## $ favourites_count : int 18066 ## $ verified : logi FALSE ## $ statuses_count : int 5545 ## $ profile_image_url_https: chr \u0026quot;https://pbs.twimg.com/profile_images/1330619806396067845/mIPmR-x4_normal.jpg\u0026quot; ## $ profile_banner_url : chr \u0026quot;https://pbs.twimg.com/profile_banners/750892662224453632/1602417664\u0026quot; ## $ default_profile : logi FALSE ## $ default_profile_image : logi FALSE ## $ withheld_in_countries :List of 1 ## ..$ : list() ## $ entities :List of 1 ## ..$ :List of 2 ## - attr(*, \u0026quot;tweets\u0026quot;)='data.frame':\t1 obs. of 36 variables: ## ..$ created_at : chr \u0026quot;Thu Jul 29 16:15:52 +0000 2021\u0026quot; ## ..$ id : num 1.42e+18 ## ..$ id_str : chr \u0026quot;1420780122949529601\u0026quot; ## ..$ text : chr \u0026quot;Looking forward to digging into this paper \\U0001f929\\U0001f9ee\\U0001f60d https://t.co/ddCMraM3kj\u0026quot; ## ..$ truncated : logi FALSE ## ..$ entities :List of 1 ## ..$ source : chr \u0026quot;\u0026lt;a href=\\\u0026quot;http://twitter.com/download/iphone\\\u0026quot; rel=\\\u0026quot;nofollow\\\u0026quot;\u0026gt;Twitter for iPhone\u0026lt;/a\u0026gt;\u0026quot; ## ..$ in_reply_to_status_id : logi NA ## ..$ in_reply_to_status_id_str : logi NA ## ..$ in_reply_to_user_id : logi NA ## ..$ in_reply_to_user_id_str : logi NA ## ..$ in_reply_to_screen_name : logi NA ## ..$ geo : logi NA ## ..$ coordinates :List of 1 ## ..$ place :List of 1 ## ..$ contributors : logi NA ## ..$ is_quote_status : logi TRUE ## ..$ quoted_status_id : num 1.42e+18 ## ..$ quoted_status_id_str : chr \u0026quot;1420768369620312065\u0026quot; ## ..$ retweet_count : int 0 ## ..$ favorite_count : int 8 ## ..$ favorited : logi FALSE ## ..$ retweeted : logi FALSE ## ..$ possibly_sensitive : logi FALSE ## ..$ lang : chr \u0026quot;en\u0026quot; ## ..$ quoted_status :List of 1 ## ..$ display_text_width : int 70 ## ..$ user :List of 1 ## ..$ full_text : logi NA ## ..$ favorited_by : logi NA ## ..$ display_text_range : logi NA ## ..$ retweeted_status : logi NA ## ..$ quoted_status_permalink : logi NA ## ..$ metadata : logi NA ## ..$ query : logi NA ## ..$ possibly_sensitive_appealable: logi NA Now I obtain the id of my followers using get_followers.\nfollowers \u0026lt;- get_followers(user = \u0026quot;oaggimenez\u0026quot;, n = og$followers_count, retryonratelimit = T) From their id, I can get the same details I got on my account using lookup_users. This function is not vectorized, therefore I use a loop. Takes some time so I saved the results and load them.\ndetails_followers \u0026lt;- NULL for (i in 1:length(followers$user_id)){ tmp \u0026lt;- try(lookup_users(followers$user_id[i], retryonratelimit = TRUE), silent = TRUE) if (length(tmp) == 1){ next } else { tmp$listed_count \u0026lt;- NULL # get rid of this column which raised some format issues, we do not it anyway details_followers \u0026lt;- bind_rows(details_followers, tmp) } } save(details_followers, file = \u0026quot;details_followers.RData\u0026quot;) What info do we have?\nload(\u0026quot;dat/details_followers.RData\u0026quot;) names(details_followers) ## [1] \u0026quot;id\u0026quot; \u0026quot;id_str\u0026quot; ## [3] \u0026quot;name\u0026quot; \u0026quot;screen_name\u0026quot; ## [5] \u0026quot;location\u0026quot; \u0026quot;description\u0026quot; ## [7] \u0026quot;url\u0026quot; \u0026quot;protected\u0026quot; ## [9] \u0026quot;followers_count\u0026quot; \u0026quot;friends_count\u0026quot; ## [11] \u0026quot;created_at\u0026quot; \u0026quot;favourites_count\u0026quot; ## [13] \u0026quot;verified\u0026quot; \u0026quot;statuses_count\u0026quot; ## [15] \u0026quot;profile_image_url_https\u0026quot; \u0026quot;profile_banner_url\u0026quot; ## [17] \u0026quot;default_profile\u0026quot; \u0026quot;default_profile_image\u0026quot; ## [19] \u0026quot;withheld_in_countries\u0026quot; \u0026quot;entities\u0026quot; In more details.\nstr(details_followers, max = 1) ## 'data.frame':\t5025 obs. of 20 variables: ## $ id : num 1.03e+18 1.42e+18 1.32e+09 1.42e+18 1.36e+18 ... ## $ id_str : chr \u0026quot;1026053171024736258\u0026quot; \u0026quot;1419513064697712640\u0026quot; \u0026quot;1317282258\u0026quot; \u0026quot;1419222504367927296\u0026quot; ... ## $ name : chr \u0026quot;Dr. Zoe Nhleko\u0026quot; \u0026quot;akash Shelke\u0026quot; \u0026quot;Rilwan Ugbedeojo\u0026quot; \u0026quot;Thaana Van Dessel\u0026quot; ... ## $ screen_name : chr \u0026quot;ZoeNhleko\u0026quot; \u0026quot;akashSh88945929\u0026quot; \u0026quot;abuhrilwan\u0026quot; \u0026quot;ThaanaD\u0026quot; ... ## $ location : chr \u0026quot;United States\u0026quot; \u0026quot;\u0026quot; \u0026quot;Lagos, Nigeria\u0026quot; \u0026quot;France\u0026quot; ... ## $ description : chr \u0026quot;Wildlife ecologist with experience on African large mammals. PhD from @UF. Looking for job opportunities in eco\u0026quot;| __truncated__ \u0026quot;\u0026quot; \u0026quot;MSc, Plant Ecology | Seeking PhD Position in Community Ecology | Civic Leader | YALI RLC \u0026amp; African Presidential\u0026quot;| __truncated__ \u0026quot;MSc student majoring in Animal Ecology @WURanimal. \\nFocused on Conservation Behavior \u0026amp; Human-Wildlife Coexistence.\u0026quot; ... ## $ url : chr \u0026quot;https://t.co/uaFY2L8BnL\u0026quot; NA NA NA ... ## $ protected : logi FALSE FALSE FALSE FALSE FALSE FALSE ... ## $ followers_count : int 2878 1 244 0 2856 4 112 6 69 476 ... ## $ friends_count : int 1325 223 2030 14 2575 99 344 32 73 1671 ... ## $ created_at : chr \u0026quot;Sun Aug 05 10:31:53 +0000 2018\u0026quot; \u0026quot;Mon Jul 26 04:21:18 +0000 2021\u0026quot; \u0026quot;Sat Mar 30 22:35:27 +0000 2013\u0026quot; \u0026quot;Sun Jul 25 09:06:36 +0000 2021\u0026quot; ... ## $ favourites_count : int 8003 0 1843 1 12360 0 484 0 83 713 ... ## $ verified : logi FALSE FALSE FALSE FALSE FALSE FALSE ... ## $ statuses_count : int 7261 0 670 1 2112 0 141 0 30 926 ... ## $ profile_image_url_https: chr \u0026quot;https://pbs.twimg.com/profile_images/1306353853571280896/rk9qAxoa_normal.jpg\u0026quot; \u0026quot;https://pbs.twimg.com/profile_images/1419513134142738433/erR2XRwf_normal.png\u0026quot; \u0026quot;https://pbs.twimg.com/profile_images/1413368052012433413/EyzbYfWB_normal.jpg\u0026quot; \u0026quot;https://pbs.twimg.com/profile_images/1419222962566217731/TQT96npR_normal.jpg\u0026quot; ... ## $ profile_banner_url : chr \u0026quot;https://pbs.twimg.com/profile_banners/1026053171024736258/1533465589\u0026quot; NA \u0026quot;https://pbs.twimg.com/profile_banners/1317282258/1624183687\u0026quot; \u0026quot;https://pbs.twimg.com/profile_banners/1419222504367927296/1627205254\u0026quot; ... ## $ default_profile : logi TRUE TRUE FALSE TRUE TRUE TRUE ... ## $ default_profile_image : logi FALSE FALSE FALSE FALSE FALSE FALSE ... ## $ withheld_in_countries :List of 5025 ## $ entities :List of 5025 ## - attr(*, \u0026quot;tweets\u0026quot;)='data.frame':\t1 obs. of 36 variables: Data exploration and visualisation Let\u0026rsquo;s display the 100 bigger accounts that follow me. First thing I learned. It is humbling to be followed by influential individuals I admire like @MicrobiomDigest, @nathanpsmad, @FrancoisTaddei, @freakonometrics, @allison_horst, @HugePossum, @apreshill and scientific journals, institutions and societies.\ndetails_followers %\u0026gt;% arrange(-followers_count) %\u0026gt;% select(screen_name, followers_count, friends_count, favourites_count) %\u0026gt;% head(n = 100) ## screen_name followers_count friends_count favourites_count ## 1 dianefrancis1 230560 132635 70 ## 2 simongerman600 218444 209489 56976 ## 3 WildlifeMag 191613 23995 14697 ## 4 cagrimbakirci 170164 9401 21334 ## 5 BernardMarr 130900 29532 15764 ## 6 ZotovMax 119625 133610 894 ## 7 _atanas_ 116717 74266 102899 ## 8 MicrobiomDigest 105825 33779 98199 ## 9 Eliances 97488 94488 15164 ## 10 PLOSBiology 89263 6646 4278 ## 11 IPBES 78086 11797 53957 ## 12 7wData 76131 84155 74609 ## 13 biconnections 63802 52918 91970 ## 14 aydintufekci43 60212 22627 99427 ## 15 AliceVachet 56381 2320 182435 ## 16 Pr_Logos 52482 50520 19852 ## 17 DrJoeNyangon 51296 20858 1413 ## 18 derekantoncich 48823 51449 347 ## 19 Afro_Herper 45215 3392 117693 ## 20 KumarAGarg 43461 47795 4052 ## 21 GatelyMark 41500 37944 1153 ## 22 overleaf 41306 7830 35289 ## 23 figshare 40610 38361 10395 ## 24 mlamons1 40415 29364 2498 ## 25 coraman 40398 2169 16548 ## 26 BritishEcolSoc 37394 1143 11685 ## 27 nathanpsmad 36760 5312 43729 ## 28 WildlifeRescue 34843 7976 61284 ## 29 rudyagovic 34063 28147 3708 ## 30 grp_resilience 34011 7998 3559 ## 31 JNakev 33427 28532 29442 ## 32 FrancoisTaddei 32631 17903 23894 ## 33 RunEducator 31248 19848 17522 ## 34 LarryLLapin 30782 34265 501 ## 35 _Alex_Iaco_ 30775 26720 2390 ## 36 owasow 30562 12932 36909 ## 37 JEcology 30534 689 3312 ## 38 valmasdel 28972 4512 4334 ## 39 freakonometrics 28038 15112 31106 ## 40 MethodsEcolEvol 26207 10347 2083 ## 41 mmw_lmw 26014 12157 1068 ## 42 drmichellelarue 25424 6281 56120 ## 43 nicolasberrod 25258 6042 12990 ## 44 DD_NaNa_ 24253 11051 39916 ## 45 callin_bull 24106 2831 10572 ## 46 INEE_CNRS 23483 1873 5089 ## 47 Datascience__ 22793 10695 216 ## 48 VisualPersist 22779 5599 79872 ## 49 coywolfassoc 22203 19491 9470 ## 50 RBGSydney 20259 16049 13435 ## 51 allison_horst 19289 2865 7725 ## 52 rhskraus 18429 20029 3503 ## 53 fred_sallee 18144 18995 3423 ## 54 thembauk 17748 2066 3260 ## 55 parolesdhist 17257 7136 7497 ## 56 CREAF_ecologia 16654 3201 11899 ## 57 ChamboncelLea 16609 4059 9797 ## 58 HugePossum 16461 12732 7806 ## 59 ScientistFemale 16379 9039 2079 ## 60 sophie_e_hill 16292 5811 10315 ## 61 mohamma64508589 16130 14550 24584 ## 62 SamuelHayat 15999 3068 9767 ## 63 MathieuAvanzi 15890 1604 8526 ## 64 Erky321 15859 16284 4841 ## 65 ConLetters 15482 5547 1489 ## 66 oceansresearch 14969 1759 7599 ## 67 WCSNewsroom 14926 11224 7277 ## 68 CMastication 14687 8091 31769 ## 69 apreshill 14106 1607 32806 ## 70 SumanMalla10 13872 11977 10946 ## 71 umontpellier 13667 1360 8745 ## 72 WriteNThrive 13615 13523 7631 ## 73 abdirashidmd 13586 2389 18417 ## 74 JackKinross 13392 11263 774 ## 75 PaulREhrlich 13383 14208 1 ## 76 teppofelin 13330 12410 264 ## 77 ConservOptimism 13249 4306 15648 ## 78 jepeteso 13137 13595 3420 ## 79 gwintrob 12484 8751 43479 ## 80 SallyR_ISLHE 12411 12477 28924 ## 81 ActuIAFr 12321 8856 2495 ## 82 Booksmag 12282 1785 332 ## 83 jusoneoftoomany 12212 11049 2296 ## 84 galaxyproject 12208 8757 813 ## 85 thepostdoctoral 12186 12802 814 ## 86 FrontMarineSci 11498 5733 5832 ## 87 jobRxiv 11442 872 786 ## 88 EcographyJourna 11381 1465 4938 ## 89 davidghamilton1 11242 2231 16244 ## 90 5amStartups 11170 12370 147062 ## 91 RosmarieKatrin 11079 11159 4060 ## 92 ScienceAndMaps 11043 8679 3596 ## 93 joelgombin 11012 10335 10868 ## 94 SteveLenhert 10990 8328 669 ## 95 Naturevolve 10973 11349 7500 ## 96 cookinforpeace 10655 2861 17519 ## 97 DrPaulEvans1 10537 7927 3131 ## 98 Ibycter 10517 4450 47359 ## 99 Sciencecomptoir 10436 1268 15800 ## 100 SteveAtLogikk 10421 8864 13123 Where do my followers come from? Second thing I learnt. Followers are folks from France (my home), USA, UK, Canada, Australia, and other European countries. I was happy to realize that I have followers from other parts of the world too, in India, South Africa, Brazil, Peru, Mexico, Nepal, Argentina, Kenya, Bangladesh, etc. and from Greece (my other home).\ndetails_followers %\u0026gt;% mutate(location = str_replace(location, \u0026quot;.*France.*\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*England\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*United Kingdom\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*London\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*Scotland\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*Montpellier\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*Canada, BC.*\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*New Caledonia\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*Marseille\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*Grenoble\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*Perú\u0026quot;, \u0026quot;Peru\u0026quot;), location = str_replace(location, \u0026quot;.*Martinique\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*La Rochelle\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*Grenoble, Rhône-Alpes\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*Avignon\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*Paris\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*paris\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*france\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*Bordeaux\u0026quot;, \u0026quot;France\u0026quot;), location = str_replace(location, \u0026quot;.*Mytilene, Lesvos, Greece\u0026quot;, \u0026quot;Greece\u0026quot;), location = str_replace(location, \u0026quot;.*Mytiline, Greece\u0026quot;, \u0026quot;Greece\u0026quot;), location = str_replace(location, \u0026quot;.*Germany.*\u0026quot;, \u0026quot;Germany\u0026quot;), location = str_replace(location, \u0026quot;.*Vienna, Austria\u0026quot;, \u0026quot;Austria\u0026quot;), location = str_replace(location, \u0026quot;.*Arusha, Tanzania\u0026quot;, \u0026quot;Tanzania\u0026quot;), location = str_replace(location, \u0026quot;.*Athens\u0026quot;, \u0026quot;Greece\u0026quot;), location = str_replace(location, \u0026quot;.*Bangor, ME\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*Bangor, Wales\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*Baton Rouge, LA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Lisboa, Portugal\u0026quot;, \u0026quot;Portugal\u0026quot;), location = str_replace(location, \u0026quot;.*Oslo\u0026quot;, \u0026quot;Norway\u0026quot;), location = str_replace(location, \u0026quot;.*Islamic Republic of Iran\u0026quot;, \u0026quot;Iran\u0026quot;), location = str_replace(location, \u0026quot;.*Madrid, España\u0026quot;, \u0026quot;Spain\u0026quot;), location = str_replace(location, \u0026quot;.*Madrid\u0026quot;, \u0026quot;Spain\u0026quot;), location = str_replace(location, \u0026quot;.*Madrid, Spain\u0026quot;, \u0026quot;Spain\u0026quot;), location = str_replace(location, \u0026quot;.*Rome, Italy\u0026quot;, \u0026quot;Italy\u0026quot;), location = str_replace(location, \u0026quot;.*Santiago, Chile\u0026quot;, \u0026quot;Chile\u0026quot;), location = str_replace(location, \u0026quot;.*Belém, Brazil\u0026quot;, \u0026quot;Brazil\u0026quot;), location = str_replace(location, \u0026quot;.*Belo Horizonte, Brazil\u0026quot;, \u0026quot;Brazil\u0026quot;), location = str_replace(location, \u0026quot;.*Berlin\u0026quot;, \u0026quot;Germany\u0026quot;), location = str_replace(location, \u0026quot;.*Mexico, ME\u0026quot;, \u0026quot;Mexico\u0026quot;), location = str_replace(location, \u0026quot;.*Mexico City\u0026quot;, \u0026quot;Mexico\u0026quot;), location = str_replace(location, \u0026quot;.*Santiago, Chile\u0026quot;, \u0026quot;Chile\u0026quot;), location = str_replace(location, \u0026quot;.*Ontario\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Aberdeen\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*Cambridge\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*Bologna, Emilia Romagna\u0026quot;, \u0026quot;Italy\u0026quot;), location = str_replace(location, \u0026quot;.*Bogotá, D.C., Colombia\u0026quot;, \u0026quot;Colombia\u0026quot;), location = str_replace(location, \u0026quot;.*Medellín, Colombia\u0026quot;, \u0026quot;Colombia\u0026quot;), location = str_replace(location, \u0026quot;.*Bruxelles, Belgique\u0026quot;, \u0026quot;Belgium\u0026quot;), location = str_replace(location, \u0026quot;.*Brasil\u0026quot;, \u0026quot;Brazil\u0026quot;), location = str_replace(location, \u0026quot;.*Budweis, Czech Republic\u0026quot;, \u0026quot;Czech Republic\u0026quot;), location = str_replace(location, \u0026quot;.*Calgary, Alberta\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Canada, BC\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Cardiff, Wales\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*Brisbane\u0026quot;, \u0026quot;Australia\u0026quot;), location = str_replace(location, \u0026quot;.*Sydney\u0026quot;, \u0026quot;Australia\u0026quot;), location = str_replace(location, \u0026quot;.*Queensland\u0026quot;, \u0026quot;Australia\u0026quot;), location = str_replace(location, \u0026quot;.*Australia\u0026quot;, \u0026quot;Australia\u0026quot;), location = str_replace(location, \u0026quot;.*Germany\u0026quot;, \u0026quot;Germany\u0026quot;), location = str_replace(location, \u0026quot;.*Vancouver\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Ottawa, Ontario\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Québec, Canada\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Winnipeg, Manitoba\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*New South Wales\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Victoria\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*British Columbia\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Norway\u0026quot;, \u0026quot;Norway\u0026quot;), location = str_replace(location, \u0026quot;.*Finland\u0026quot;, \u0026quot;Finland\u0026quot;), location = str_replace(location, \u0026quot;.*South Africa\u0026quot;, \u0026quot;South Africa\u0026quot;), location = str_replace(location, \u0026quot;.*Switzerland\u0026quot;, \u0026quot;Switzerland\u0026quot;), location = str_replace(location, \u0026quot;.*CO\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*OK\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*KS\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*MS\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*CO\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*CO\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*CO\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*CO\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*WA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*MD\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Colorado\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Community of Valencia, Spain\u0026quot;, \u0026quot;Spain\u0026quot;), location = str_replace(location, \u0026quot;.*Dunedin City, New Zealand\u0026quot;, \u0026quot;New Zealand\u0026quot;), location = str_replace(location, \u0026quot;.*Rio Claro, Brasil\u0026quot;, \u0026quot;Brazil\u0026quot;), location = str_replace(location, \u0026quot;.*Saskatoon, Saskatchewan\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Sherbrooke, Québec\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*United Kingdom.*\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*University of Oxford\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*University of St Andrews\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*ID\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*NE\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*United States of America, USA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Spain, Spain\u0026quot;, \u0026quot;Spain\u0026quot;), location = str_replace(location, \u0026quot;.*USA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Wisconsin, USA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Florida, USA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Liege, Belgium\u0026quot;, \u0026quot;Belgium\u0026quot;), location = str_replace(location, \u0026quot;.*Ghent, Belgium\u0026quot;, \u0026quot;Belgium\u0026quot;), location = str_replace(location, \u0026quot;.*Pune, India\u0026quot;, \u0026quot;India\u0026quot;), location = str_replace(location, \u0026quot;.*Hyderabad, India\u0026quot;, \u0026quot;India\u0026quot;), location = str_replace(location, \u0026quot;.*Prague, Czech Republic\u0026quot;, \u0026quot;Czech Republic\u0026quot;), location = str_replace(location, \u0026quot;.*Canada, BC, Canada\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*CA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*United States.*\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*DC\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*FL\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*GA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*HI\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*ME\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*MA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*MI\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*PA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*NC\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*MO\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*NY\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*NH\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*IL\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*NM\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*MT\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*OR\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*WY\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*WI\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*MN\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*CT\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*TX\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*VA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*OH\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Massachusetts, USA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*California, USA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Montréal, Québec\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Edmonton, Alberta\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Toronto, Ontario\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Canada, Canada\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Montreal\u0026quot;, \u0026quot;Canada\u0026quot;), location = str_replace(location, \u0026quot;.*Lisbon, Portugal\u0026quot;, \u0026quot;Portugal\u0026quot;), location = str_replace(location, \u0026quot;.*Coimbra, Portugal\u0026quot;, \u0026quot;Portugal\u0026quot;), location = str_replace(location, \u0026quot;.*Cork, Ireland\u0026quot;, \u0026quot;Ireland\u0026quot;), location = str_replace(location, \u0026quot;.*Dublin City, Ireland\u0026quot;, \u0026quot;Ireland\u0026quot;), location = str_replace(location, \u0026quot;.*Barcelona, Spain\u0026quot;, \u0026quot;Spain\u0026quot;), location = str_replace(location, \u0026quot;.*Barcelona\u0026quot;, \u0026quot;Spain\u0026quot;), location = str_replace(location, \u0026quot;.*Leipzig\u0026quot;, \u0026quot;Germany\u0026quot;), location = str_replace(location, \u0026quot;.*Seville, Spain\u0026quot;, \u0026quot;Spain\u0026quot;), location = str_replace(location, \u0026quot;.*Seville, Spain\u0026quot;, \u0026quot;Spain\u0026quot;), location = str_replace(location, \u0026quot;.*Buenos Aires, Argentina\u0026quot;, \u0026quot;Argentina\u0026quot;), location = str_replace(location, \u0026quot;.*Rio de Janeiro, Brazil\u0026quot;, \u0026quot;Brazil\u0026quot;), location = str_replace(location, \u0026quot;.*Canberra\u0026quot;, \u0026quot;Australia\u0026quot;), location = str_remove(location, \u0026quot;Global\u0026quot;), location = str_remove(location, \u0026quot;Earth\u0026quot;), location = str_remove(location, \u0026quot;Worldwide\u0026quot;), location = str_remove(location, \u0026quot;Europe\u0026quot;), location = str_remove(location, \u0026quot; \u0026quot;), location = str_replace(location, \u0026quot;.*Dhaka, Bangladesh\u0026quot;, \u0026quot;Bangladesh\u0026quot;), location = str_replace(location, \u0026quot;.*Copenhagen, Denmark\u0026quot;, \u0026quot;Denmark\u0026quot;), location = str_replace(location, \u0026quot;.*Amsterdam, The Netherlands\u0026quot;, \u0026quot;The Netherlands\u0026quot;), location = str_replace(location, \u0026quot;.*Groningen, Nederland\u0026quot;, \u0026quot;The Netherlands\u0026quot;), location = str_replace(location, \u0026quot;.*Wageningen, Nederland\u0026quot;, \u0026quot;The Netherlands\u0026quot;), location = str_replace(location, \u0026quot;.*Aarhus, Denmark\u0026quot;, \u0026quot;Denmark\u0026quot;), location = str_replace(location, \u0026quot;.*Antwerp, Belgium\u0026quot;, \u0026quot;Belgium\u0026quot;), location = str_replace(location, \u0026quot;.*Aveiro, Portugal\u0026quot;, \u0026quot;Portugal\u0026quot;), location = str_replace(location, \u0026quot;.*Australia, AUS\u0026quot;, \u0026quot;Australia\u0026quot;), location = str_replace(location, \u0026quot;.*Australian National University\u0026quot;, \u0026quot;Australia\u0026quot;), location = str_replace(location, \u0026quot;.*Auckland, New Zealand\u0026quot;, \u0026quot;New Zealand\u0026quot;), location = str_replace(location, \u0026quot;.*Belfast, Northern Ireland\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*Ireland\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*Hobart, Tasmania\u0026quot;, \u0026quot;Australia\u0026quot;), location = str_replace(location, \u0026quot;.*Dhaka, Bangladesh\u0026quot;, \u0026quot;Bangladesh\u0026quot;), location = str_replace(location, \u0026quot;.*Nairobi, Kenya\u0026quot;, \u0026quot;Kenya\u0026quot;), location = str_replace(location, \u0026quot;.*Dhaka, Bangladesh\u0026quot;, \u0026quot;Bangladesh\u0026quot;), location = str_replace(location, \u0026quot;.*Berlin, Deutschland\u0026quot;, \u0026quot;Germany\u0026quot;), location = str_replace(location, \u0026quot;.*Munich, Bavaria\u0026quot;, \u0026quot;Germany\u0026quot;), location = str_replace(location, \u0026quot;.*Dehradun, India\u0026quot;, \u0026quot;India\u0026quot;), location = str_replace(location, \u0026quot;.*Bengaluru, India\u0026quot;, \u0026quot;India\u0026quot;), location = str_replace(location, \u0026quot;.*Berlin, Deutschland\u0026quot;, \u0026quot;Germany\u0026quot;), location = str_replace(location, \u0026quot;.*Deutschland\u0026quot;, \u0026quot;Germany\u0026quot;), location = str_replace(location, \u0026quot;.*Edinburgh\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*Glasgow\u0026quot;, \u0026quot;United Kingdom\u0026quot;), location = str_replace(location, \u0026quot;.*New York, USA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Washington, USA\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*California\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*California\u0026quot;, \u0026quot;United States of America\u0026quot;), location = str_replace(location, \u0026quot;.*Christchurch City, New Zealand\u0026quot;, \u0026quot;New Zealand\u0026quot;), location = str_replace(location, \u0026quot;.*Harare, Zimbabwe\u0026quot;, \u0026quot;Zimbabwe\u0026quot;), location = str_replace(location, \u0026quot;.*Islamabad, Pakistan\u0026quot;, \u0026quot;Pakistan\u0026quot;), location = str_replace(location, \u0026quot;.*Kolkata, India\u0026quot;, \u0026quot;India\u0026quot;), location = str_replace(location, \u0026quot;.*Lagos, Nigeria\u0026quot;, \u0026quot;Nigeria\u0026quot;), location = str_replace(location, \u0026quot;.*Lima, Peru\u0026quot;, \u0026quot;Peru\u0026quot;), location = str_replace(location, \u0026quot;.*Valparaíso, Chile\u0026quot;, \u0026quot;Chile\u0026quot;), location = str_replace(location, \u0026quot;.*Uppsala, Sweden\u0026quot;, \u0026quot;Sweden\u0026quot;), location = str_replace(location, \u0026quot;.*Uppsala, Sverige\u0026quot;, \u0026quot;Sweden\u0026quot;), location = str_replace(location, \u0026quot;.*Stockholm, Sweden\u0026quot;, \u0026quot;Sweden\u0026quot;), location = str_replace(location, \u0026quot;.*Stockholm\u0026quot;, \u0026quot;Sweden\u0026quot;), location = str_replace(location, \u0026quot;.*Turin, Piedmont\u0026quot;, \u0026quot;Italy\u0026quot;), location = str_replace(location, \u0026quot;.*University of Iceland\u0026quot;, \u0026quot;Iceland\u0026quot;), location = str_replace(location, \u0026quot;.*University of Helsinki\u0026quot;, \u0026quot;Finland\u0026quot;), location = str_replace(location, \u0026quot;.*Tucumán, Argentina\u0026quot;, \u0026quot;Argentina\u0026quot;), location = str_replace(location, \u0026quot;.*The Hague, The Netherlands\u0026quot;, \u0026quot;The Netherlands\u0026quot;), location = str_replace(location, \u0026quot;.*UK\u0026quot;, \u0026quot;United Kingdom\u0026quot;) ) %\u0026gt;% count(location, sort = TRUE) %\u0026gt;% #slice(-40) %\u0026gt;% head(n = 45) -\u0026gt; followers_by_country followers_by_country ## location n ## 1 1273 ## 2 United States of America 606 ## 3 France 571 ## 4 United Kingdom 473 ## 5 Canada 166 ## 6 Germany 130 ## 7 Australia 112 ## 8 Norway 50 ## 9 Switzerland 43 ## 10 Spain 34 ## 11 India 33 ## 12 South Africa 32 ## 13 Brazil 30 ## 14 Portugal 27 ## 15 Finland 26 ## 16 Sweden 18 ## 17 Belgium 17 ## 18 Italy 16 ## 19 The Netherlands 14 ## 20 New Zealand 13 ## 21 Peru 13 ## 22 Denmark 11 ## 23 Mexico 11 ## 24 Nepal 10 ## 25 Argentina 9 ## 26 Kenya 9 ## 27 Bangladesh 8 ## 28 Chile 8 ## 29 Greece 8 ## 30 Colombia 6 ## 31 Czech Republic 6 ## 32 Singapore 6 ## 33 Iceland 5 ## 34 Senegal 5 ## 35 Iran 4 ## 36 Nigeria 4 ## 37 Pakistan 4 ## 38 Tanzania 4 ## 39 Zimbabwe 4 ## 40 3 ## 41 Austria 3 ## 42 Ecuador 3 ## 43 Guatemala 3 ## 44 Hong Kong 3 ## 45 Luxembourg 3 Let\u0026rsquo;s have a look to the same info on a quick and dirty map. White countries do not appear in the 45 countries with most followers, or do not have any followers.\nlibrary(rworldmap) library(classInt) spdf \u0026lt;- joinCountryData2Map(followers_by_country, joinCode=\u0026quot;NAME\u0026quot;, nameJoinColumn=\u0026quot;location\u0026quot;, verbose=TRUE) ## 43 codes from your data successfully matched countries in the map ## 2 codes from your data failed to match with a country code in the map ## failedCodes failedCountries ## [1,] \u0026quot;\u0026quot; \u0026quot;\u0026quot; ## [2,] \u0026quot;\u0026quot; \u0026quot; \u0026quot; ## 200 codes from the map weren't represented in your data classInt \u0026lt;- classIntervals(spdf$n, n=9, style = \u0026quot;jenks\u0026quot;) catMethod \u0026lt;- classInt[[\u0026quot;brks\u0026quot;]] library(RColorBrewer) colourPalette \u0026lt;- brewer.pal(9,'RdPu') mapParams \u0026lt;- mapCountryData(spdf, nameColumnToPlot=\u0026quot;n\u0026quot;, addLegend=FALSE, catMethod = catMethod, colourPalette=colourPalette, mapTitle=\u0026quot;Number of followers per country\u0026quot;) do.call(addMapLegend, c(mapParams, legendLabels=\u0026quot;all\u0026quot;, legendWidth=0.5, legendIntervals=\u0026quot;data\u0026quot;, legendMar = 2)) How many followers are verified?\ndetails_followers %\u0026gt;% count(verified) ## verified n ## 1 FALSE 5009 ## 2 TRUE 16 Who are the verified users? Scientific entities, but also some scientists.\ndetails_followers %\u0026gt;% filter(verified==TRUE) %\u0026gt;% select(name, screen_name, location, followers_count) %\u0026gt;% arrange(-followers_count) ## name screen_name ## 1 Çağrı Mert Bakırcı cagrimbakirci ## 2 Elisabeth Bik MicrobiomDigest ## 3 ipbes IPBES ## 4 Dr. Earyn McGee, Lizard lassoer \\U0001f98e Afro_Herper ## 5 figshare figshare ## 6 Nathan Peiffer-Smadja nathanpsmad ## 7 Wildlife Alliance WildlifeRescue ## 8 Arthur Charpentier freakonometrics ## 9 L'écologie au CNRS INEE_CNRS ## 10 The Royal Botanic Garden Sydney RBGSydney ## 11 CREAF CREAF_ecologia ## 12 Jean-Paul Carrera JeanCarrPaul ## 13 Deckset decksetapp ## 14 𝙳𝚞𝚌-𝚀𝚞𝚊𝚗𝚐 𝙽𝚐𝚞𝚢𝚎𝚗 duc_qn ## 15 Christophe Josset \\U0001f4dd CJosset ## 16 Casey Fung TheCaseyFung ## location followers_count ## 1 Texas, USA 170164 ## 2 San Francisco, CA 105825 ## 3 Bonn, Germany 78086 ## 4 United States 45215 ## 5 London 40610 ## 6 Paris - London 36760 ## 7 Phnom Penh, Cambodia 34843 ## 8 Montréal, Québec 28038 ## 9 23483 ## 10 Sydney, New South Wales 20259 ## 11 Bellaterra (Barcelona) 16654 ## 12 Oxford, England 5744 ## 13 Your Mac 3884 ## 14 Lausanne, Switzerland 2808 ## 15 Paris, France 2002 ## 16 Bundjalung Country | Mullum 1612 How old are my followers, where age is defined as the time elapsed since they created their Twitter account. Twitter was created in 2006. I created my account in July 2016 but started using it only two or three years ago I\u0026rsquo;d say. This dataviz is not too much informative I guess.\ndetails_followers %\u0026gt;% mutate(month = created_at %\u0026gt;% str_split(\u0026quot; \u0026quot;) %\u0026gt;% map_chr(2), day = created_at %\u0026gt;% str_split(\u0026quot; \u0026quot;) %\u0026gt;% map_chr(3), year = created_at %\u0026gt;% str_split(\u0026quot; \u0026quot;) %\u0026gt;% map_chr(6), created_at = paste0(month,\u0026quot;-\u0026quot;,year), created_at = lubridate::my(created_at)) %\u0026gt;% group_by(created_at) %\u0026gt;% summarise(counts = n()) %\u0026gt;% ggplot(aes(x = created_at, y = counts)) + geom_line() + labs(x = \u0026quot;date of creation\u0026quot;, y = \u0026quot;how many accounts were created this month-year\u0026quot;) What is the distribution of the number of followers and friends. On average, 400 followers and 400 friends.\nfoll \u0026lt;- details_followers %\u0026gt;% ggplot() + aes(x = log(followers_count)) + geom_histogram(fill=\u0026quot;#69b3a2\u0026quot;, color=\u0026quot;#e9ecef\u0026quot;, alpha=0.9) + labs(x = \u0026quot;log number of followers\u0026quot;, y = \u0026quot;counts\u0026quot;, title = \u0026quot;followers\u0026quot;) friends \u0026lt;- details_followers %\u0026gt;% ggplot() + aes(x = log(friends_count)) + geom_histogram(fill=\u0026quot;#69b3a2\u0026quot;, color=\u0026quot;#e9ecef\u0026quot;, alpha=0.9) + labs(x = \u0026quot;log number of friends\u0026quot;, y = \u0026quot;\u0026quot;, title = \u0026quot;friends\u0026quot;) library(patchwork) foll + friends Retrieving social links To investigate the social network made of my followers, we need to gather who they follow, and their followers. Unfortunately, the Twitter API imposes some time and quantity constraints in retrieving data.\nHere, for the sake of illustration, I will focus on some accounts only. To do so, I consider accounts with between 1000 and 1500 followers, and more than 30000 tweets users have liked in their account\u0026rsquo;s lifetime (which denotes some activity). The more followers we consider, the more time it would take to gather info on them: it is approx one minute per follower on average, so one hour for sixty followers, and more than eighty hours for all my followers! With the filters I\u0026rsquo;m using, I end up with 35 followers here, this is very few to say infer meaningful about a network, but it\u0026rsquo;ll do the job for now.\nsome_followers \u0026lt;- details_followers %\u0026gt;% filter((followers_count \u0026gt; 1000 \u0026amp; followers_count \u0026lt; 1500), favourites_count \u0026gt; 30000) # %\u0026gt;% nrow() Create empty list and name it after their screen name.\nfoler \u0026lt;- vector(mode = 'list', length = length(some_followers$screen_name)) names(foler) \u0026lt;- some_followers$screen_name Get followers of these selected followers. Takes ages, so save and load later.\nfor (i in 1:length(some_followers$screen_name)) { message(\u0026quot;Getting followers for user #\u0026quot;, i, \u0026quot;/\u0026quot;, nrow(some_followers)) foler[[i]] \u0026lt;- get_followers(some_followers$screen_name[i], n = some_followers$followers_count[i], retryonratelimit = TRUE) if(i %% 5 == 0){ message(\u0026quot;sleep for 5 minutes\u0026quot;) Sys.sleep(5*60) } } save(foler, file = \u0026quot;foler.RData\u0026quot;) Format the data.\nload(\u0026quot;dat/foler.RData\u0026quot;) folerx \u0026lt;- bind_rows(foler, .id = \u0026quot;screen_name\u0026quot;) active_fol_x \u0026lt;- some_followers %\u0026gt;% select(id_str, screen_name) foler_join \u0026lt;- left_join(folerx, some_followers, by = \u0026quot;screen_name\u0026quot;) algo_follower \u0026lt;- foler_join %\u0026gt;% select(id_str, screen_name) %\u0026gt;% setNames(c(\u0026quot;follower\u0026quot;, \u0026quot;active_user\u0026quot;)) %\u0026gt;% na.omit() Get friends of my followers. Takes ages. Again I save and load later.\nfriend \u0026lt;- data.frame() for (i in seq_along(some_followers$screen_name)) { message(\u0026quot;Getting following for user #\u0026quot;, i ,\u0026quot;/\u0026quot;,nrow(some_followers)) kk \u0026lt;- get_friends(some_followers$screen_name[i], n = some_followers$friends_count[i], retryonratelimit = TRUE) friend \u0026lt;- rbind(friend, kk) if(i %% 15 == 0){ message(\u0026quot;sleep for 15 minutes\u0026quot;) Sys.sleep(15*60+1) } } save(friend, file = \u0026quot;friend.RData\u0026quot;) Format the data.\nload(\u0026quot;dat/friend.RData\u0026quot;) all_friend \u0026lt;- friend %\u0026gt;% setNames(c(\u0026quot;screen_name\u0026quot;, \u0026quot;user_id\u0026quot;)) all_friendx \u0026lt;- left_join(all_friend, active_fol_x, by=\u0026quot;screen_name\u0026quot;) algo_friend \u0026lt;- all_friendx %\u0026gt;% select(user_id, screen_name) %\u0026gt;% setNames(c(\u0026quot;following\u0026quot;,\u0026quot;active_user\u0026quot;)) Now that we have all info on followers and friends, we\u0026rsquo;re gonna build the network of people who follow each other.\nun_active \u0026lt;- unique(algo_friend$active_user) %\u0026gt;% data.frame(stringsAsFactors = F) %\u0026gt;% setNames(\u0026quot;active_user\u0026quot;) algo_mutual \u0026lt;- data.frame() for (i in seq_along(un_active$active_user)){ aa \u0026lt;- algo_friend %\u0026gt;% filter(active_user == un_active$active_user[i]) bb \u0026lt;- aa %\u0026gt;% filter(aa$following %in% algo_follower$follower) %\u0026gt;% setNames(c(\u0026quot;mutual\u0026quot;,\u0026quot;active_user\u0026quot;)) algo_mutual \u0026lt;- rbind(algo_mutual,bb) } Instead of ids, we use screen names.\ndetail_friend \u0026lt;- lookup_users(algo_mutual$mutual) algo_mutual \u0026lt;- algo_mutual %\u0026gt;% left_join(detail_friend, by = c(\u0026quot;mutual\u0026quot; = \u0026quot;id_str\u0026quot;)) %\u0026gt;% na.omit() %\u0026gt;% select(mutual, active_user, screen_name) algo_mutual ## # A tibble: 18 × 3 ## mutual active_user screen_name ## \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;chr\u0026gt; ## 1 824265743575306243 g33k5p34k robbie_emmet ## 2 761735501707567104 g33k5p34k LauraBlissEco ## 3 1439272374 alexnicolharper LeafyEricScott ## 4 824265743575306243 alexnicolharper robbie_emmet ## 5 21467726 alexnicolharper g33k5p34k ## 6 4181949437 LegalizeBrain EvpokPadding ## 7 824265743575306243 Souzam7139 robbie_emmet ## 8 888890201673580548 Souzam7139 mellenmartin ## 9 824265743575306243 mellenmartin robbie_emmet ## 10 21467726 LauraBlissEco g33k5p34k ## 11 831398755920445440 JosiahParry RoelandtN42 ## 12 4181949437 gau EvpokPadding ## 13 21467726 robbie_emmet g33k5p34k ## 14 4159201575 robbie_emmet alexnicolharper ## 15 888890201673580548 robbie_emmet mellenmartin ## 16 29538964 EvpokPadding gau ## 17 1439272374 lifedispersing LeafyEricScott ## 18 301687349 RoelandtN42 JosiahParry Add my account to the network.\nun_active \u0026lt;- un_active %\u0026gt;% mutate(mutual = rep(\u0026quot;oaggimenez\u0026quot;)) un_active \u0026lt;- un_active[,c(2,1)] un_active \u0026lt;- un_active %\u0026gt;% setNames(c(\u0026quot;active_user\u0026quot;,\u0026quot;screen_name\u0026quot;)) algo_mutual \u0026lt;- bind_rows(algo_mutual %\u0026gt;% select(-mutual), un_active) For what follows, we will need packages to work with networks.\nlibrary(igraph) library(tidygraph) library(ggraph) Now we create the edges, nodes and build the network.\nnodes \u0026lt;- data.frame(V = unique(c(algo_mutual$screen_name,algo_mutual$active_user)), stringsAsFactors = F) edges \u0026lt;- algo_mutual %\u0026gt;% setNames(c(\u0026quot;from\u0026quot;,\u0026quot;to\u0026quot;)) network_ego1 \u0026lt;- graph_from_data_frame(d = edges, vertices = nodes, directed = F) %\u0026gt;% as_tbl_graph() Network metrics Create communities using group_louvain() algorithm, and calculate standard metrics using tidygraph.\nset.seed(123) network_ego1 \u0026lt;- network_ego1 %\u0026gt;% activate(nodes) %\u0026gt;% mutate(community = as.factor(group_louvain())) %\u0026gt;% mutate(degree_c = centrality_degree()) %\u0026gt;% mutate(betweenness_c = centrality_betweenness(directed = F,normalized = T)) %\u0026gt;% mutate(closeness_c = centrality_closeness(normalized = T)) %\u0026gt;% mutate(eigen = centrality_eigen(directed = F)) network_ego_df \u0026lt;- as.data.frame(network_ego1) Identify key nodes with respect to network metrics.\nnetwork_ego_df ## name community degree_c betweenness_c closeness_c eigen ## 1 robbie_emmet 2 8 0.0049299720 0.5384615 0.5238676 ## 2 LauraBlissEco 2 3 0.0000000000 0.5147059 0.2717415 ## 3 LeafyEricScott 5 3 0.0008403361 0.5223881 0.2287709 ## 4 g33k5p34k 2 6 0.0024649860 0.5303030 0.4342056 ## 5 EvpokPadding 3 4 0.0011204482 0.5223881 0.2336052 ## 6 mellenmartin 2 4 0.0000000000 0.5223881 0.3371889 ## 7 RoelandtN42 4 3 0.0000000000 0.5147059 0.2050991 ## 8 alexnicolharper 2 5 0.0019607843 0.5303030 0.3942456 ## 9 gau 3 3 0.0000000000 0.5147059 0.2133909 ## 10 JosiahParry 4 3 0.0000000000 0.5147059 0.2050991 ## 11 genius_c137 1 1 0.0000000000 0.5072464 0.1454399 ## 12 Zjbb 1 1 0.0000000000 0.5072464 0.1454399 ## 13 SJRAfloat 1 1 0.0000000000 0.5072464 0.1454399 ## 14 BMPARMA17622540 1 1 0.0000000000 0.5072464 0.1454399 ## 15 maryam_adeli 1 1 0.0000000000 0.5072464 0.1454399 ## 16 waywardaf 1 1 0.0000000000 0.5072464 0.1454399 ## 17 SusyVF6 1 1 0.0000000000 0.5072464 0.1454399 ## 18 Sinalo_NM 1 1 0.0000000000 0.5072464 0.1454399 ## 19 Leila_Lula 1 1 0.0000000000 0.5072464 0.1454399 ## 20 CARThorpe 1 1 0.0000000000 0.5072464 0.1454399 ## 21 j_wilson_white 1 1 0.0000000000 0.5072464 0.1454399 ## 22 LegalizeBrain 3 2 0.0000000000 0.5147059 0.1794154 ## 23 JaishriJuice 1 1 0.0000000000 0.5072464 0.1454399 ## 24 jaguaretepy 1 1 0.0000000000 0.5072464 0.1454399 ## 25 Souzam7139 2 3 0.0000000000 0.5223881 0.2706719 ## 26 LaLince_ 1 1 0.0000000000 0.5072464 0.1454399 ## 27 Nina_Ella_ 1 1 0.0000000000 0.5072464 0.1454399 ## 28 samcox 1 1 0.0000000000 0.5072464 0.1454399 ## 29 TAdamsBio42 1 1 0.0000000000 0.5072464 0.1454399 ## 30 brubakerl 1 1 0.0000000000 0.5072464 0.1454399 ## 31 robanhk 1 1 0.0000000000 0.5072464 0.1454399 ## 32 InesCCarv 1 1 0.0000000000 0.5072464 0.1454399 ## 33 groundhog0202 1 1 0.0000000000 0.5072464 0.1454399 ## 34 zenmart 1 1 0.0000000000 0.5072464 0.1454399 ## 35 lifedispersing 5 2 0.0000000000 0.5147059 0.1787123 ## 36 oaggimenez 1 35 0.9685154062 1.0000000 1.0000000 Get users with highest values of each metric.\nkp_ego \u0026lt;- data.frame( network_ego_df %\u0026gt;% arrange(-degree_c) %\u0026gt;% select(name), network_ego_df %\u0026gt;% arrange(-betweenness_c) %\u0026gt;% select(name), network_ego_df %\u0026gt;% arrange(-closeness_c) %\u0026gt;% select(name), network_ego_df %\u0026gt;% arrange(-eigen) %\u0026gt;% select(name)) %\u0026gt;% setNames(c(\u0026quot;degree\u0026quot;,\u0026quot;betweenness\u0026quot;,\u0026quot;closeness\u0026quot;,\u0026quot;eigen\u0026quot;)) kp_ego[-1,] ## degree betweenness closeness eigen ## 2 robbie_emmet robbie_emmet robbie_emmet robbie_emmet ## 3 g33k5p34k g33k5p34k g33k5p34k g33k5p34k ## 4 alexnicolharper alexnicolharper alexnicolharper alexnicolharper ## 5 EvpokPadding EvpokPadding LeafyEricScott mellenmartin ## 6 mellenmartin LeafyEricScott EvpokPadding LauraBlissEco ## 7 LauraBlissEco LauraBlissEco mellenmartin Souzam7139 ## 8 LeafyEricScott mellenmartin Souzam7139 EvpokPadding ## 9 RoelandtN42 RoelandtN42 LauraBlissEco LeafyEricScott ## 10 gau gau RoelandtN42 gau ## 11 JosiahParry JosiahParry gau RoelandtN42 ## 12 Souzam7139 genius_c137 JosiahParry JosiahParry ## 13 LegalizeBrain Zjbb LegalizeBrain LegalizeBrain ## 14 lifedispersing SJRAfloat lifedispersing lifedispersing ## 15 genius_c137 BMPARMA17622540 genius_c137 robanhk ## 16 Zjbb maryam_adeli Zjbb Sinalo_NM ## 17 SJRAfloat waywardaf SJRAfloat samcox ## 18 BMPARMA17622540 SusyVF6 BMPARMA17622540 genius_c137 ## 19 maryam_adeli Sinalo_NM maryam_adeli Zjbb ## 20 waywardaf Leila_Lula waywardaf SJRAfloat ## 21 SusyVF6 CARThorpe SusyVF6 BMPARMA17622540 ## 22 Sinalo_NM j_wilson_white Sinalo_NM waywardaf ## 23 Leila_Lula LegalizeBrain Leila_Lula SusyVF6 ## 24 CARThorpe JaishriJuice CARThorpe Leila_Lula ## 25 j_wilson_white jaguaretepy j_wilson_white j_wilson_white ## 26 JaishriJuice Souzam7139 JaishriJuice JaishriJuice ## 27 jaguaretepy LaLince_ jaguaretepy jaguaretepy ## 28 LaLince_ Nina_Ella_ LaLince_ LaLince_ ## 29 Nina_Ella_ samcox Nina_Ella_ Nina_Ella_ ## 30 samcox TAdamsBio42 samcox brubakerl ## 31 TAdamsBio42 brubakerl TAdamsBio42 groundhog0202 ## 32 brubakerl robanhk brubakerl zenmart ## 33 robanhk InesCCarv robanhk maryam_adeli ## 34 InesCCarv groundhog0202 InesCCarv CARThorpe ## 35 groundhog0202 zenmart groundhog0202 InesCCarv ## 36 zenmart lifedispersing zenmart TAdamsBio42 Robbie Emmet has the highest degree, betweenness, closeness centrality and eigenvector centrality. He has the most relations with the other nodes in the network. He also can spread information the further away and faster than anyone, and is also surrounded by important persons in the network. This is within the network we\u0026rsquo;ve just built.\nBy the way, Robbie has just passed his dissertation defense, congrats! Follow him on Twitter at @robbie_emmet, he is awesome!\nVisualize Network We visualize the network using clusters or communities (first three only).\noptions(ggrepel.max.overlaps = Inf) network_viz \u0026lt;- network_ego1 %\u0026gt;% filter(community %in% 1:3) %\u0026gt;% mutate(node_size = ifelse(degree_c \u0026gt;= 50,degree_c,0)) %\u0026gt;% mutate(node_label = ifelse(betweenness_c \u0026gt;= 0.01,name,NA)) plot_ego \u0026lt;- network_viz %\u0026gt;% ggraph(layout = \u0026quot;stress\u0026quot;) + geom_edge_fan(alpha = 0.05) + geom_node_point(aes(color = as.factor(community),size = node_size)) + geom_node_label(aes(label = node_label),nudge_y = 0.1, show.legend = F, fontface = \u0026quot;bold\u0026quot;, fill = \u0026quot;#ffffff66\u0026quot;) + theme_graph() + theme(legend.position = \u0026quot;none\u0026quot;) + labs(title = \u0026quot;Top 3 communities\u0026quot;) plot_ego Let\u0026rsquo;s use an alternative dataviz to identify the nodes.\nnetwork_viz %\u0026gt;% mutate(community = group_spinglass()) %\u0026gt;% ggraph(layout = \u0026quot;nicely\u0026quot;) + geom_edge_fan(alpha = 0.25) + geom_node_point(aes(color = factor(community)),size = 5, show.legend = F) + geom_node_text(aes(label = name),repel = T) + theme_graph() + theme(legend.position = \u0026quot;none\u0026quot;) + labs(title = \u0026quot;Network with named nodes\u0026quot;) Concluding words Not sure what the communities are about. Due to limitation with the Twitter API I did not attempt to retrieve the info on all the 5,000 followers, but I used only 35 of them, which explains why there is not much to say about this network I guess. I might take some time to try and retrieve all the info (will take days), or not. I might also use some information I already have, like how followers describe themselves and do some topic modelling to identify common themes of interest.\nHere I have simply followed the steps that Joe Cristian illustrated so brillantly in his post at https://algotech.netlify.app/blog/social-network-analysis-in-r/. Make sure you read his post for more details on the code and theory. In particular, he illustrates how information spreads through the network.\nFor more about social network analyses and Twitter, see this thread https://twitter.com/Mehdi_Moussaid/status/1389174715990876160?s=20 and this video https://www.youtube.com/watch?v=UX7YQ6m2r_o in French, this is what inspired me to do the analysis. See also this https://github.com/eleurent/twitter-graph for a much better analysis than what I could do (with Python for data retrieving and manipulation, and Gephi for network visualisation).\nThis post was also published on https://www.r-bloggers.com.\u0026#160;\u0026#x21a9;\u0026#xfe0e;\n","date":1627430400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1627430400,"objectID":"40f652687c71880ce5f02374b6de0deb","permalink":"https://oliviergimenez.github.io/blog/twitter-social-network/","publishdate":"2021-07-28T00:00:00Z","relpermalink":"/blog/twitter-social-network/","section":"blog","summary":"I use R to retrieve some data from Twitter, do some exploratory data analysis and visualisation and examine a network of followers.\n","tags":["Twitter","social network","R","rstats"],"title":"Quick and dirty analysis of a Twitter social network","type":"blog"},{"authors":null,"categories":null,"content":"It was awesome!\nBig day today 🥳 Hélène Labach is defending her PhD on the conservation of Bottlenose 🐬 in the French Mediterranean 🌊 @MIRACETI_ceta @cefemontpellier @umontpellier @CollegeDoctoUM @twitthair1 pic.twitter.com/RkbHSe415Q\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 24, 2021 ","date":1624492800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1624492800,"objectID":"052faa6ce386aa5343660bba1182d1a7","permalink":"https://oliviergimenez.github.io/blog/lnphddefense/","publishdate":"2021-06-24T00:00:00Z","relpermalink":"/blog/lnphddefense/","section":"blog","summary":"It was awesome!\n","tags":["population ecology","Conservation biology"],"title":"Hélène Labach's PhD defense on the conservation of Bottlenose dolphins in the French Mediterranean sea","type":"blog"},{"authors":null,"categories":null,"content":"Slides and codes available from https://sarahbauduin.github.io/formation_IBM_NetLogoR/.\nOur workshop on individual-based models #IBMs with NetLogoR #rstats starts today, led by #twitterless IBM expert Sarah Bauduin @OFBiodiversite @cefemontpellier w/ @oksanagrente \u0026amp; @NSantostasi\n▶️ Slides \u0026amp; codes https://t.co/LjCh4AH0UA\n▶️ NetLogoR https://t.co/KXQxIGwKMh pic.twitter.com/MjvVZxMi7H\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 21, 2021 ","date":1624320000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1624320000,"objectID":"2c5acfbe92fe4bfe53b206237f9d55a8","permalink":"https://oliviergimenez.github.io/blog/ibmworkshop/","publishdate":"2021-06-22T00:00:00Z","relpermalink":"/blog/ibmworkshop/","section":"blog","summary":"Slides and codes available from https://sarahbauduin.github.io/formation_IBM_NetLogoR/.\n","tags":["rstats","individual-based models"],"title":"Workshop on individual-based models with R","type":"blog"},{"authors":null,"categories":null,"content":"Our poster with Ruth King and Rachel McCrea for the 2021 Euring conference.\nPoster session at @_EURING #euring2021 feels like a video game from back in the days @gather_town 🥳 Sadly, couldn\u0026#39;t stay as I was on 🧇 duty 😋 Btw our poster w/ @BayesRuth \u0026amp; @RSMcCrea is a review of the capture-recapture literature over the last decade, check out ⤵️ pic.twitter.com/B2Gal2Sj6I\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 8, 2021 ","date":1623110400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1623110400,"objectID":"c0985876325312cf08f9afcc1bad8262","permalink":"https://oliviergimenez.github.io/blog/euring2021/","publishdate":"2021-06-08T00:00:00Z","relpermalink":"/blog/euring2021/","section":"blog","summary":"Our poster with Ruth King and Rachel McCrea for the 2021 Euring conference.\n","tags":["Capture-recapture","Literature review","Conference","Poster"],"title":"A review of the capture-recapture literature over the last decade","type":"blog"},{"authors":null,"categories":null,"content":"Journée de restitution sur notre projet ERC-Lynx.\nLancement de la journée de restitution de notre projet #ittecop \u0026quot;Eviter, réduire et compenser la mortalité du Lynx par collision avec les véhicules de transport\u0026quot; @ademe @FRBiodiv @Ecologie_Gouv Plus sur le projet ici https://t.co/uKY5ylOli2 pic.twitter.com/e0TdSqvhUZ\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 4, 2021 ","date":1622764800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1622764800,"objectID":"5912b711a934ee57ac239c8bc5277898","permalink":"https://oliviergimenez.github.io/blog/restitutionerc/","publishdate":"2021-06-04T00:00:00Z","relpermalink":"/blog/restitutionerc/","section":"blog","summary":"Journée de restitution sur notre projet ERC-Lynx.\n","tags":["Lynx lynx","Car-wildlife collisions","Human-Wildlife conflicts"],"title":"Journée de restitution du projet ERC Lynx","type":"blog"},{"authors":null,"categories":null,"content":"Everything (including data, slides, codes and video recordings) about the workshop we run on Bayesian capture-recapture inference with hidden Markov models in R and Nimble.\n🥳🍻 Had fun running a workshop on Bayesian capture-recapture inference with hidden Markov models in R #rstats \u0026amp; Nimble w/ @SarahCubaynes @chloe_nater @MaudQueroue \u0026amp; Perry de Valpine @R_nimble ✅ Slides, worksheets \u0026amp; video recordings freely available from https://t.co/CgVd84MPYN pic.twitter.com/WlHzXytVUG\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) May 18, 2021 ","date":1621296000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1621296000,"objectID":"7f3925442bafb377695f4d37d73b3dac","permalink":"https://oliviergimenez.github.io/blog/hmmcrworkshop/","publishdate":"2021-05-18T00:00:00Z","relpermalink":"/blog/hmmcrworkshop/","section":"blog","summary":"Everything (including data, slides, codes and video recordings) about the workshop we run on Bayesian capture-recapture inference with hidden Markov models in R and Nimble.\n","tags":["Bayesian statistics","Capture-recapture models","rstats","Nimble"],"title":"Workshop on Bayesian capture-recapture inference","type":"blog"},{"authors":null,"categories":null,"content":"Everything (including data, slides and codes) about the workshop on Bayesian statistics I run.\nJust finished up a week teaching Bayesian statistics with R #rstats for non-stats PhD students \u0026amp; 💜 it #bayes #mcmc #jags\n🌍 Dedicated website https://t.co/X7JY3f9eGX\n🍳 Material https://t.co/agY9CvQLiV\n📽️ Video recording (in English w/ 👽 accent) https://t.co/DHabXmVYsT pic.twitter.com/g94nSO3faQ\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 26, 2021 ","date":1616716800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1616716800,"objectID":"e7a5270e0c95180e8190f0abce1d80d6","permalink":"https://oliviergimenez.github.io/blog/bayesstatsworkshop/","publishdate":"2021-03-26T00:00:00Z","relpermalink":"/blog/bayesstatsworkshop/","section":"blog","summary":"Everything (including data, slides and codes) about the workshop on Bayesian statistics I run.\n","tags":["Bayesian statistics","rstats"],"title":"Workshop on Bayesian statistics with R","type":"blog"},{"authors":null,"categories":null,"content":"Everything (including data, slides and codes) about the workshop on reproducible science I run.\nHad fun running a 1-day workshop on reproducible science in R \u0026amp; RStudio for our lab @cefemontpellier. We covered data manip \u0026amp; visualisation w/ #tidyverse, version control w/ Git/GitHub and writing dynamic docs w/ R Markdown\n📝 Slides \u0026amp; practicals https://t.co/LJRknr8Kzb 📽️=🇫🇷 pic.twitter.com/uCsm9kH0Hf\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 21, 2021 ","date":1616284800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1616284800,"objectID":"53cfcbee95df7f8b1c2d6d03f2858f60","permalink":"https://oliviergimenez.github.io/blog/rrworkshop/","publishdate":"2021-03-21T00:00:00Z","relpermalink":"/blog/rrworkshop/","section":"blog","summary":"Everything (including data, slides and codes) about the workshop on reproducible science I run.\n","tags":["Reproducible science","rstats","RStudio"],"title":"Workshop on reproducible science in R \u0026 RStudio","type":"blog"},{"authors":null,"categories":null,"content":"More about our project on mitigating collisions between lynx and cars.\n📽️🎙️ La vidéo et les diapos du webinaire sont disponibles https://t.co/xt2EXsvIGN - 🙏 aux participant.e.s pour leurs questions, et à l\u0026#39;équipe #Ittecop pour l\u0026#39;organisation ! https://t.co/Ix7Zkse02H\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 8, 2021 ","date":1615161600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1615161600,"objectID":"01a1a74013cd423dcfb3aa2f0678a497","permalink":"https://oliviergimenez.github.io/blog/commerc/","publishdate":"2021-03-08T00:00:00Z","relpermalink":"/blog/commerc/","section":"blog","summary":"More about our project on mitigating collisions between lynx and cars.\n","tags":["Human-wildlife conflicts","Lynx lynx","Car-wildlife collisions","SciComm"],"title":"Communication sur notre projet ERC-Lynx","type":"blog"},{"authors":null,"categories":null,"content":"Our take on deep learning, computer vision and ecology.\n\u0026quot;Images, écologie et deep learning\u0026quot; co-écrit avec Vincent Miele et Stéphane Dray @LbbeLyon sur les applications du #DeepLearning à l\u0026#39;analyse d\u0026#39;images en écologie et difficultés associées. 🙏 @sfecologie pour l\u0026#39;opportunité https://t.co/0cTYbHTb35 @INEE_CNRS @cefemontpellier\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 22, 2021 ","date":1613952000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1613952000,"objectID":"59d0320da40df81624533f20d2410842","permalink":"https://oliviergimenez.github.io/blog/sfeeia/","publishdate":"2021-02-22T00:00:00Z","relpermalink":"/blog/sfeeia/","section":"blog","summary":"Our take on deep learning, computer vision and ecology.\n","tags":["Artificial intelligence","Deep learning","Computer vision","GDR EcoStat"],"title":"Regard SFEE Images, écologie et deep learning","type":"blog"},{"authors":null,"categories":null,"content":"Some highlights on the year 2020 for our group.\nOur research group had a fantastic 2020 year! A thread w/ some highlights 🥳🍾 https://t.co/BuNSAceisM 0/9 pic.twitter.com/uPuAjV5uve\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 23, 2020 ","date":1608768000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1608768000,"objectID":"b93a279462a5c9f64e26813b903746b6","permalink":"https://oliviergimenez.github.io/blog/retrospective2020/","publishdate":"2020-12-24T00:00:00Z","relpermalink":"/blog/retrospective2020/","section":"blog","summary":"Some highlights on the year 2020 for our group.\n","tags":["laughter","retrospective","research group"],"title":"We had a great 2020 year, considering...","type":"blog"},{"authors":null,"categories":null,"content":"Workshop to come on reproducible science in our lab.\n🧑‍🔬💻♻️ Planning a 1-day workshop on reproducible science for our lab @cefemontpellier. We\u0026#39;ll use R and RStudio, Git, GitHub, R Markdown, tidyverse and sf 🤩\nMaterial here https://t.co/QApKQsl1FU #rstats #ReproducibleScience @twitthair1 pix by @scriberian pic.twitter.com/LFBwZmelit\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 18, 2020 ","date":1608249600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1608249600,"objectID":"10792b7b86f34b3580cdb7e37ae7eee8","permalink":"https://oliviergimenez.github.io/blog/reproscience2020/","publishdate":"2020-12-18T00:00:00Z","relpermalink":"/blog/reproscience2020/","section":"blog","summary":"Workshop to come on reproducible science in our lab.\n","tags":["reproducible science","R","rstat","git","GitHub","tidyverse"],"title":"Workshop on reproducible science","type":"blog"},{"authors":null,"categories":null,"content":"A seminar I gave remotely for the university of Lisboa.\nStatistical ecology 🌍📊, large carnivores 🐺🐻 and a pinch of social sciences 🤷, happy to share our research in this 45\u0026#39; talk 🤩🍾\nslides: https://t.co/fKTMWko5J9\nvideo: https://t.co/7KubqYJIBU pic.twitter.com/OJ292VWuel\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 10, 2020 ","date":1607558400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1607558400,"objectID":"f3bc928c9dd62acad9b683a7f3a6a5c8","permalink":"https://oliviergimenez.github.io/blog/talklisboa/","publishdate":"2020-12-10T00:00:00Z","relpermalink":"/blog/talklisboa/","section":"blog","summary":"A seminar I gave remotely for the university of Lisboa.\n","tags":["large carnivores","statistical ecology","seminar"],"title":"Seminar on large carnivores and statistical ecology","type":"blog"},{"authors":null,"categories":null,"content":"Just finished a workshop on integrated population models.\nAnd it\u0026#39;s oooooooooooon ! https://t.co/sEy70RCORN pic.twitter.com/nNBCvIonjV\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) November 23, 2020 ","date":1606089600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1606089600,"objectID":"a5701df93c48a04007f484b65fa379e7","permalink":"https://oliviergimenez.github.io/blog/ipmworkshop_thread/","publishdate":"2020-11-23T00:00:00Z","relpermalink":"/blog/ipmworkshop_thread/","section":"blog","summary":"Just finished a workshop on integrated population models.\n","tags":["integrated population models","capture-recapture","workshop"],"title":"Workshop on integrated population models","type":"blog"},{"authors":null,"categories":null,"content":"Co-organisé les journées annuelles Mate-shs pendant lesquelles on a parlé des nouveaux métiers et outils autour des données en SHS.\nLes journées annuelles Mate-shs se tiennent aujourd\u0026#39;hui et demain 🎉🤩 On y parle des nouveaux métiers et outils autour des données en SHS @mate_shs_cnrs @MSHsud @CNRS @INSHS_CNRS Site web : https://t.co/oDJsi9deUo\nProgramme : https://t.co/kKkklKhEsd#SHS #interdisciplinarité pic.twitter.com/bliBnZm8qY\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) November 19, 2020 ","date":1605744000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1605744000,"objectID":"c525b990816a0ac4f3cee8ab5016fd3d","permalink":"https://oliviergimenez.github.io/blog/mateshs/","publishdate":"2020-11-19T00:00:00Z","relpermalink":"/blog/mateshs/","section":"blog","summary":"Co-organisé les journées annuelles Mate-shs pendant lesquelles on a parlé des nouveaux métiers et outils autour des données en SHS.\n","tags":["SHS","science des données","sciences sociales","journées"],"title":"Journées annuelles réseau Mate-SHS","type":"blog"},{"authors":null,"categories":null,"content":"Co-organised a workshop on deep learning applications in ecology. Thread below and my notes there.\n🥁🤩 Our workshop on #DeepLearning and #Ecology is about to begin. We\u0026#39;ll be talking about how to process and analyse images 📸 and sounds 🎙️ Program: https://t.co/gUXD0tMCW1\nCurated list of resources for ecologists: https://t.co/HieQfqgVEx GDR EcoStat: https://t.co/kTXpjJXlya pic.twitter.com/nfjVke9ETU\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) November 16, 2020 ","date":1605484800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1605484800,"objectID":"469f2c2df4be9c6aa3893dd54ca1b03a","permalink":"https://oliviergimenez.github.io/blog/deeplearning/","publishdate":"2020-11-16T00:00:00Z","relpermalink":"/blog/deeplearning/","section":"blog","summary":"Co-organised a workshop on deep learning applications in ecology. Thread below and my notes there.\n","tags":["deep learning","workshops","computer vision","statistical ecology"],"title":"Workshop on Deep Learning and Ecology","type":"blog"},{"authors":null,"categories":null,"content":"The material (slides and video) of my course on statistics for ecologists for Master students.\nEvery year I teach a 12h course on statistics for ecologists for Master students at @umontpellier covering freq/bayes GLM(M)s. The material of the course is available ➡️ https://t.co/Y1CTvkmH8Q w/ slides, practicals \u0026amp; videos (in French).#StatisticalEcology #ReproducibleTeaching pic.twitter.com/A15vndiw49\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) November 12, 2020 ","date":1605139200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1605139200,"objectID":"a207a18f3cc4f797b6446e2e54202ce1","permalink":"https://oliviergimenez.github.io/blog/stat-course-2020/","publishdate":"2020-11-12T00:00:00Z","relpermalink":"/blog/stat-course-2020/","section":"blog","summary":"The material (slides and video) of my course on statistics for ecologists for Master students.\n","tags":["statistical ecology","Master course"],"title":"Master course on statistics for ecologists","type":"blog"},{"authors":null,"categories":null,"content":"Notre article pour la Fête de la Science dans The Conversation \u0026ldquo; Mieux connaître le lynx boréal grâce à l\u0026rsquo;écologie statistique\u0026rdquo;.\nJetez un coup d\u0026#39;oeil à notre article pour la Fête de la Science dans @FR_Conversation \u0026quot;Mieux connaître le lynx boréal grâce à l\u0026#39;écologie statistique\u0026quot; https://t.co/d40TzNcmis co-écrit avec @LouvrierJulie \u0026amp; Sarah Bauduin - avec l\u0026#39;aide précieuse de Françoise Marmouyet #FDS2020 pic.twitter.com/xmJX0crmmy\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) October 8, 2020 ","date":1602115200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1602115200,"objectID":"da6e1b0a4adaea0943753d6a9fd5ea4a","permalink":"https://oliviergimenez.github.io/blog/theconversation/","publishdate":"2020-10-08T00:00:00Z","relpermalink":"/blog/theconversation/","section":"blog","summary":"Notre article pour la Fête de la Science dans The Conversation \u0026ldquo; Mieux connaître le lynx boréal grâce à l\u0026rsquo;écologie statistique\u0026rdquo;.\n","tags":["The Conversation","écologie statistique","lynx boréal","scicomm"],"title":"Article The Conversation sur le lynx et l'écologie statistique","type":"blog"},{"authors":null,"categories":null,"content":"Gathered some code on occupancy, capture-recapture \u0026amp; epidemiological models, social networks, spatial stuff, textual analyses, reproducible science, etc\u0026hellip;\n👩‍💻👨‍💻 In hope it\u0026#39;s useful, I gathered random pieces of code I wrote on occupancy, capture-recapture \u0026amp; epidemiological models, social networks, spatial stuff, textual analyses, #DeepLearning and the #tidyverse ➡️ https://t.co/T1PlbmDLZO I also went full purple @MoonbeamLevels 💜 pic.twitter.com/q6vK4LYVS7\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) August 7, 2020 ","date":1596758400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1596758400,"objectID":"d19a6bfefff5762e590b101a802dcb2d","permalink":"https://oliviergimenez.github.io/blog/pieceofcode/","publishdate":"2020-08-07T00:00:00Z","relpermalink":"/blog/pieceofcode/","section":"blog","summary":"Gathered some code on occupancy, capture-recapture \u0026amp; epidemiological models, social networks, spatial stuff, textual analyses, reproducible science, etc\u0026hellip;\n","tags":["R","Nimble","Jags","coding"],"title":"Some random piece of code","type":"blog"},{"authors":null,"categories":["project"],"content":"Large carnivores are difficult to monitor in the field: they are elusive and live at low densities over wide areas. With their recent recovery in Europe, large carnivores are having increasing interactions with human activities. We contribute to the estimation of their abundance and distribution by applying and developing methods that make optimal use of available data. We are also working on applying science-based management using the structured decision making framework. Past and ongoing projects were are are mostly led by Master students, PhDs and post-docs.\n","date":1596672000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1596672000,"objectID":"3d31a390c54fc2c129512091d2cf0878","permalink":"https://oliviergimenez.github.io/my-project/external-project/carnivores/","publishdate":"2020-08-06T00:00:00Z","relpermalink":"/my-project/external-project/carnivores/","section":"my-project","summary":"Developing statistical tools to survey large carnivores in Europe.","tags":["carnivores","monitoring","management"],"title":"Large carnivores monitoring","type":"my-project"},{"authors":null,"categories":null,"content":"","date":1596672000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1596672000,"objectID":"dc306a4b5877ae8802b6579f50075fd7","permalink":"https://oliviergimenez.github.io/my-project/external-project/lynx/","publishdate":"2020-08-06T00:00:00Z","relpermalink":"/my-project/external-project/lynx/","section":"my-project","summary":"Avoiding, reducing and compensating mortality risks of the Eurasian Lynx by collision with transport vehicles.","tags":["carnivores","collisions","management","software"],"title":"Terrestrial transport infrastructures, Lynx population viability and mitigation strategies","type":"my-project"},{"authors":null,"categories":null,"content":"","date":1595203200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1595203200,"objectID":"077e1e01fd8590c86aff39607410dbcb","permalink":"https://oliviergimenez.github.io/project_landing/","publishdate":"2020-07-20T00:00:00Z","relpermalink":"/project_landing/","section":"","summary":"See some of the projects I have worked on.","tags":null,"title":"Projects","type":"widget_page"},{"authors":null,"categories":null,"content":"Trained in mathematics and statistics in particular, I have developed a bottomless taste for questions in statistical ecology.\nI often resort to state-space models, and hidden Markov models (e.g., here and here) in particular, to develop statistical methods for capture-recapture, occupancy and integrated population models.\nIn our group, we address questions in conservation biology to assess species viability, in ecology to study the effect of climate change on animal demography, in evolution to examine life-history tactics and individual heterogeneity and in wildlife management to develop strategies for conservation and conflict mitigation. To account for the human dimension inherent to these questions, I went back to the university to study sociology. A nice illustration of this interdisciplinary work is Gilles Maurer\u0026rsquo;s PhD combining anthropology, demography, economics and genetics to better understand the interactions between captive and wild population of Asian elephants.\nI\u0026rsquo;m doing my best to teach statistics (including the Bayesian way) to the students of the Ecology and Evolution Master in Montpellier. We also make efforts to structure the community in France through our national research group in statistical ecology. Last but not least, every now and then, our group runs workshops to diffuse quantitative methods in the ecological community.\n","date":1595203200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1595203200,"objectID":"363851c9ba77641b8792e2edf5dc8a5c","permalink":"https://oliviergimenez.github.io/my-project/external-project/methods/","publishdate":"2020-07-20T00:00:00Z","relpermalink":"/my-project/external-project/methods/","section":"my-project","summary":"Developing, communicating and teaching methods and models for ecology and conservation biology.","tags":["methods","capture-recapture","R","Bayes","MCMC"],"title":"Statistical ecology","type":"my-project"},{"authors":null,"categories":null,"content":"Just finished polishing up my slides for a workshop on Bayesian statistics.\nJust finished polishing up my slides for a workshop on Bayesian statistics 🥱😅 ☑️ Slides, code \u0026amp; data https://t.co/1SrGkYO8o0\n☑️ Videos 🎦 soon to follow (in French, then English w/ strong accent).#rstats #jags #BayesianStatistics #MCMC #EcologicalStatistics #MultilevelModels pic.twitter.com/lxleExWjqE\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 18, 2020 ","date":1595030400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1595030400,"objectID":"3ecb62d28f23fbd957c234767fac2809","permalink":"https://oliviergimenez.github.io/blog/bayesworkshop/","publishdate":"2020-07-18T00:00:00Z","relpermalink":"/blog/bayesworkshop/","section":"blog","summary":"Just finished polishing up my slides for a workshop on Bayesian statistics.\n","tags":["Bayes","workshops","statistical ecology"],"title":"Workshop on Bayesian statistics","type":"blog"},{"authors":null,"categories":null,"content":"Quick and dirty text mining of the ISEC2020 abstracts.\nQuick and dirty text mining of the @isec2020 #vISEC2020 abstracts using awesome \u0026#39;Text Mining with R - A Tidy Approach\u0026#39; by @juliasilge \u0026amp; @drob https://t.co/qwugbZevDY - codes \u0026amp; data ➡️ https://t.co/Y4adeWUpiW pic.twitter.com/bPjAgNwsWL\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 5, 220 ","date":1593993600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1593993600,"objectID":"8c08a9ad386aedfbe20564764c16e7f7","permalink":"https://oliviergimenez.github.io/blog/textminingisec/","publishdate":"2020-07-06T00:00:00Z","relpermalink":"/blog/textminingisec/","section":"blog","summary":"Quick and dirty text mining of the ISEC2020 abstracts.\n","tags":["text mining","tidyverse","isec2020"],"title":"Text mining of the ISEC2020 abstracts","type":"blog"},{"authors":null,"categories":null,"content":"Tribute to Byron Morgan\nWas visiting your website @RSMcCrea and read again the Byronmetrics paper 😍🤩 https://t.co/9XDRWagU4p that you put together back for @ISECStatEcol 2014 in Montpellier to honour Byron Morgan\u0026#39;s retirement - Thx for doing that, and say hello to Byron when you see him 😉 pic.twitter.com/E4ADQqQrBR\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 4, 220 ","date":1593820800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1593820800,"objectID":"9711fc587ab89ccb9887510f1a1a3bd5","permalink":"https://oliviergimenez.github.io/blog/byron/","publishdate":"2020-07-04T00:00:00Z","relpermalink":"/blog/byron/","section":"blog","summary":"Tribute to Byron Morgan\n","tags":["ISEC2014","Byron Morgan"],"title":"Byronmetrics","type":"blog"},{"authors":null,"categories":null,"content":"Talking about statistical ecology and large carnivores.\nTalking about statistical ecology 🌎💻📊 \u0026amp; large carnivores 🐻🐺😺 is always fun 🥳 Check out some work we\u0026#39;re doing in our @twitthair1 group https://t.co/EPZpr0FP9I, w/ guests stars 🤩@SCubi25 @LMarescot #SarahBauduin @isec2020 @ISECStatEcol @LouvrierJulie @oksanagrente pic.twitter.com/euH0TdO3Hy\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) June 2, 220 ","date":1591056000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1591056000,"objectID":"4c6c85d86a49203d142f31d104d932bb","permalink":"https://oliviergimenez.github.io/blog/publiclecture2/","publishdate":"2020-06-02T00:00:00Z","relpermalink":"/blog/publiclecture2/","section":"blog","summary":"Talking about statistical ecology and large carnivores.\n","tags":["large carnivores","public lecture","statistical ecology"],"title":"Statistical ecology and large carnivores","type":"blog"},{"authors":null,"categories":null,"content":"Tagging species with deep learning.\nTagging species w/ #Keras #Retinanet https://t.co/lZRlrDHtfM @FizyrBV Not too bad so far 🤩 green box is truth, blue/orange box is prediction #DeepLearning #CameraTraps More details ➡️ https://t.co/HieQfqgVEx pic.twitter.com/uhocKDuuGP\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) May 11, 220 ","date":1589155200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1589155200,"objectID":"173f1417293edf177fb74743fc4dc7a1","permalink":"https://oliviergimenez.github.io/blog/dlcameras/","publishdate":"2020-05-11T00:00:00Z","relpermalink":"/blog/dlcameras/","section":"blog","summary":"Tagging species with deep learning.\n","tags":["deep learning","camera traps"],"title":"Tagging species with deep learning.","type":"blog"},{"authors":null,"categories":null,"content":"Crude attempt at fitting dynamic ODE-BASED models to noisy data with Nimble.\nCrude attempt at fitting dynamic ODE-BASED models to noisy data with #nimble @R_nimble https://t.co/xmmPtSUR2R I got to call R functions in the code, change MCMC algorithms, do block sampling, \u0026amp; all in #rstats environment 🤩 #StatisticalEcology Thx D Turek \u0026amp; P de Valpine for help pic.twitter.com/MgyI1c5WLH\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 2020 ","date":1587686400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1587686400,"objectID":"42ac37c99f133ae8729f092ba099ff1e","permalink":"https://oliviergimenez.github.io/blog/odenimble/","publishdate":"2020-04-24T00:00:00Z","relpermalink":"/blog/odenimble/","section":"blog","summary":"Crude attempt at fitting dynamic ODE-BASED models to noisy data with Nimble.\n","tags":["ODE","Nimble"],"title":"Fitting ODE-based model to data in Nimble","type":"blog"},{"authors":null,"categories":null,"content":"Interactive data visualisation of bias in occupancy models w/ flexdashboard.\nInteractive #dataviz of bias in occupancy models w/ #flexdashboard (aka #shinyapps for the rest of us) https://t.co/fkYiHfAaEU #rstats @rstudio #unmarked https://t.co/KKGTPE8eJh pic.twitter.com/Rstpdv8CqJ\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 27,020 ","date":1587686400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1587686400,"objectID":"050895ce54c9472ccd011ae1e7c2b7ff","permalink":"https://oliviergimenez.github.io/blog/interactiveocc/","publishdate":"2020-04-24T00:00:00Z","relpermalink":"/blog/interactiveocc/","section":"blog","summary":"Interactive data visualisation of bias in occupancy models w/ flexdashboard.\n","tags":["occupancy","flexdashboard","r","shiny"],"title":"Interactive visualisation of bias in occupancy models","type":"blog"},{"authors":null,"categories":null,"content":"How to simulate data from the Rota\u0026rsquo;s multispecies occupancy model.\nI\u0026#39;m working a lot w/ multispecies occupancy models these days, needed to simulate data from models à la Rota et al. (2016), in case it is useful to others ➡️ #rstats code https://t.co/qVzaXLoYcV Feedbacks welcome 😉 See also comments ⬇️ pic.twitter.com/fDNBp48GXT\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 13, 2020 A *Bayesian* follow-up on fitting a 2-species occupancy model w/ #unmarked to #rstats simulated data: How to fit the simplest Rota\u0026#39;s model in #Jags and @R_nimble ➡️ https://t.co/BmEAx0cRvw Feedback is welcome 😊 https://t.co/PCO8pGLtLB pic.twitter.com/OpSskddTxP\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 14, 2020 ","date":1586736000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1586736000,"objectID":"b8dc6ae70c9198b95237fad1d5ba4f4d","permalink":"https://oliviergimenez.github.io/blog/rota/","publishdate":"2020-04-13T00:00:00Z","relpermalink":"/blog/rota/","section":"blog","summary":"How to simulate data from the Rota\u0026rsquo;s multispecies occupancy model.\n","tags":["occupancy","simulations","Rota"],"title":"Simulations from the Rota's multispecies occupancy model","type":"blog"},{"authors":null,"categories":null,"content":"Quick \u0026amp; dirty simulations to look at bias in occupancy estimates with respect to species detection for a static/single-season model.\nQuick \u0026amp; dirty simulations to look at bias in occupancy estimates with respect to species detection for a static/single-season model https://t.co/RRNpXuB3la #rstats #occupancy #unmarked pic.twitter.com/eN5kkdbkfm\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 12, 2020 ","date":1586649600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1586649600,"objectID":"8d1dff32845e5edf25a293bee40fe2cb","permalink":"https://oliviergimenez.github.io/blog/occusimu/","publishdate":"2020-04-12T00:00:00Z","relpermalink":"/blog/occusimu/","section":"blog","summary":"Quick \u0026amp; dirty simulations to look at bias in occupancy estimates with respect to species detection for a static/single-season model.\n","tags":["occupancy","simulations","bias"],"title":"Bias in occupancy models","type":"blog"},{"authors":null,"categories":null,"content":"Une analyse descriptive du phénomène de concentration d’attaques de loup sur des élevages d’animaux domestiques en France.\nAvec des collègues de @INRAE_France, @Supagro, #CERPAM et @OFBiodiversite, nous proposons une analyse descriptive du phénomène de concentration d’attaques de loup sur des élevages d’animaux domestiques en France https://t.co/Ueyq5gkhi1 - #rstats ➡️ https://t.co/RKtvqjBVk0 pic.twitter.com/qoup5jKauv\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 12, 2020 ","date":1586649600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1586649600,"objectID":"8efbac66c95f22a42496bd51987754e2","permalink":"https://oliviergimenez.github.io/blog/coadapht/","publishdate":"2020-04-12T00:00:00Z","relpermalink":"/blog/coadapht/","section":"blog","summary":"Une analyse descriptive du phénomène de concentration d’attaques de loup sur des élevages d’animaux domestiques en France.\n","tags":["wolf","tidyverse","large carnivores","predation"],"title":"Déprédations par le loup en France","type":"blog"},{"authors":null,"categories":null,"content":"Bayesian implementation of Pollock\u0026rsquo;s robust design capture-recapture models w/ Jags.\nBayesian implementation of Pollock\u0026#39;s @kpollock48 robust design capture-recapture models w/ #jags #rstats from a @MethodsEcolEvol paper by T. Riecke and colleagues ➡️ https://t.co/nE7N8kHPiR pic.twitter.com/ZToHY3luZR\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) April 6, 2020 ","date":1586131200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1586131200,"objectID":"dd2e9e8057e997ee48e695485b355129","permalink":"https://oliviergimenez.github.io/blog/rdcmr/","publishdate":"2020-04-06T00:00:00Z","relpermalink":"/blog/rdcmr/","section":"blog","summary":"Bayesian implementation of Pollock\u0026rsquo;s robust design capture-recapture models w/ Jags.\n","tags":["robust design","rstats","capture-recapture"],"title":"Bayesian implementation of the robust design","type":"blog"},{"authors":null,"categories":null,"content":"In our weekly group meeting this morning, I introduced R Markdown a great #rstats tool to write reproducible documents (reports, articles, slides, websites) smoothly mixing text, code, figures and equations in html, word or pdf format.\nIn our weekly @twitthair1 meeting this morning, I introduced R Markdown https://t.co/Hza76oAw4K a great #rstats tool to write reproducible documents (reports, articles, slides, websites) smoothly mixing text, code, figures and equations in html, word or pdf format. Ressources ⬇️ pic.twitter.com/r8vKbEaz04\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 31, 2020\u003c\u003e ","date":1585612800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1585612800,"objectID":"b597fb39065631e32071005001e61178","permalink":"https://oliviergimenez.github.io/blog/rmarkdown/","publishdate":"2020-03-31T00:00:00Z","relpermalink":"/blog/rmarkdown/","section":"blog","summary":"In our weekly group meeting this morning, I introduced R Markdown a great #rstats tool to write reproducible documents (reports, articles, slides, websites) smoothly mixing text, code, figures and equations in html, word or pdf format.","tags":["R Markdown","rstats","reproducible science"],"title":"R Markdown","type":"blog"},{"authors":null,"categories":null,"content":"In a recent seminar, S. Gandon (w/ S. Lion and T. Day) used SIR models to illustrate the effect of diff strategies of #SocialDistancing on the #COVID19 epidemic. What better way to learn than to reproduce their results in R? My two cents #rstats code ➡️ https://t.co/oJVQHJ6HqO 🤓 pic.twitter.com/dqSIcjbhqq\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 29, 2/a\u003e ","date":1585526400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1585526400,"objectID":"cd0613bb9d7b45e8b078285f4a5754c5","permalink":"https://oliviergimenez.github.io/blog/sirstats/","publishdate":"2020-03-30T00:00:00Z","relpermalink":"/blog/sirstats/","section":"blog","summary":"In a recent seminar, S. Gandon (w/ S. Lion and T. Day) used SIR models to illustrate the effect of diff strategies of #SocialDistancing on the #COVID19 epidemic. What better way to learn than to reproduce their results in R?","tags":["SIR","rstats","ODEs"],"title":"SIR models in R","type":"blog"},{"authors":null,"categories":null,"content":"Ongoing work on a flexible spatio-temporal model for occupancy data w/ a GAM-like approach in Nimble.\n😺📉 Ongoing work on a flexible spatio-temporal model for #occupancy data w/ GAM-like approach in @R_nimble\nThis work wouldn\u0026#39;t be possible wo/ the dedication of @OFBiodiversite colleagues 🙏💯 Slides➡️https://t.co/MSjwmtCCSq GAM anim courtesy @ucfagls\nThx @Yvan2935 for the tweet https://t.co/zsyTyh1Qru\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 11, 2/a\u003e ","date":1583884800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1583884800,"objectID":"ead821b85a3abaa1f765069cb57d7ffd","permalink":"https://oliviergimenez.github.io/blog/gamoccupancy/","publishdate":"2020-03-11T00:00:00Z","relpermalink":"/blog/gamoccupancy/","section":"blog","summary":"Ongoing work on a flexible spatio-temporal model for occupancy data w/ a GAM-like approach in Nimble.\n","tags":["occupancy","GAM","large carnivores"],"title":"GAM-like modelling of lynx occupancy in France","type":"blog"},{"authors":null,"categories":null,"content":"On my way to Rennes for our annual meeting on Statistical Ecology, looking forward to catching up w/ the French crowd.\nOn my way to @agrocampusouest for our annual meeting on #StatisticalEcology, looking forward to catching up w/ the french crowd #SemaineDesMaths #gdrecostat @Marie_Etienne @twitthair1 pic.twitter.com/2TGDer4wea\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 9, 2/a\u003e ","date":1583712000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1583712000,"objectID":"3bef4e6acdedeb0e8b5c8a22d013bc0b","permalink":"https://oliviergimenez.github.io/blog/gdrecostat/","publishdate":"2020-03-09T00:00:00Z","relpermalink":"/blog/gdrecostat/","section":"blog","summary":"On my way to Rennes for our annual meeting on Statistical Ecology, looking forward to catching up w/ the French crowd.\n","tags":["statistical ecology","Rennes"],"title":"French annual meeting in statistical ecology","type":"blog"},{"authors":null,"categories":null,"content":"It was so great to have you @ZipkinLab w/ us, it was too short though, come back soon 😋🚀 Monarch 🦋 are fascinating, data integration is the way to go! @twitthair1 pic.twitter.com/9ud1ycPZvE\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 6, 2/a\u003e ","date":1583452800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1583452800,"objectID":"ee66bd8c8a46af1501de1ec7122fa74a","permalink":"https://oliviergimenez.github.io/blog/elisevisit/","publishdate":"2020-03-06T00:00:00Z","relpermalink":"/blog/elisevisit/","section":"blog","summary":"It was so great to have you @ZipkinLab w/ us, it was too short though, come back soon 😋🚀 Monarch 🦋 are fascinating, data integration is the way to go! @twitthair1 pic.","tags":["community ecology","IPM"],"title":"Elise Zipkin's visit","type":"blog"},{"authors":null,"categories":null,"content":"It is my pleasure to announce that Michael Schaub and Marc Kéry will run their Integrated Population Modelling workshop in Montpellier this November.\nIt\u0026#39;s my pleasure to announce that #MichaelSchaub \u0026amp; #MarcKéry will run their Integrated Population Modelling #IPM workshop in Montpellier 🇫🇷 this November, with help from @MaudQueroue and I - details ➡️ https://t.co/pgCSxwwJW3. Come over, it\u0026#39;s going to be aaaaawesome 😁 pic.twitter.com/JvoXk07Jb1\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 4020 ","date":1583280000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1583280000,"objectID":"0f9717e88e30dc86693b47b56439cf55","permalink":"https://oliviergimenez.github.io/blog/ipmworkshop/","publishdate":"2020-03-04T00:00:00Z","relpermalink":"/blog/ipmworkshop/","section":"blog","summary":"It is my pleasure to announce that Michael Schaub and Marc Kéry will run their Integrated Population Modelling workshop in Montpellier this November.\n","tags":["IPM","workshop","rstats"],"title":"IPM workshop","type":"blog"},{"authors":null,"categories":null,"content":"Le 5 mars, l\u0026rsquo;université et la recherche s\u0026rsquo;arrêtent.\n✊ 5 mars, l\u0026#39;université et la recherche s\u0026#39;arrêtent - on est prêt 😍 #lppr #FacsEtLabosEnLutte #emploisESR Pour en savoir plus, écoutez ce #podcast https://t.co/ALGnXbgWPw @parolesdhist et si vous voulez expliquer à vos collègues 🇬🇧🇺🇸 https://t.co/iZXBsWlgFl @MHoussayH pic.twitter.com/xxKuDF6f1e\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 4, 2020 ","date":1583280000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1583280000,"objectID":"ad06ecfd93306ed1169fe829b6483f76","permalink":"https://oliviergimenez.github.io/blog/manif/","publishdate":"2020-03-04T00:00:00Z","relpermalink":"/blog/manif/","section":"blog","summary":"Le 5 mars, l\u0026rsquo;université et la recherche s\u0026rsquo;arrêtent.\n","tags":["ESR","demonstration"],"title":"L'université et la recherche s'arrêtent","type":"blog"},{"authors":null,"categories":null,"content":"I will soon be talking to an audience of Master students giving thoughts about how ecology and statistics may interact (aka Statistical Ecology), with large carnivors as a case study.\nI will soon be talking to an audience of Master students giving thoughts about how ecology \u0026amp; statistics may interact (aka #StatisticalEcology), w/ #LargeCarnivores as a case study - slides ➡️ https://t.co/ekCskObNaG. Thx @Marie_Etienne for the invit. Comments more than welcome! pic.twitter.com/A0lMPhJ5r1\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) March 1020 ","date":1583020800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1583020800,"objectID":"b254379c86700875ea1fc51973b674ac","permalink":"https://oliviergimenez.github.io/blog/publiclecture/","publishdate":"2020-03-01T00:00:00Z","relpermalink":"/blog/publiclecture/","section":"blog","summary":"I will soon be talking to an audience of Master students giving thoughts about how ecology and statistics may interact (aka Statistical Ecology), with large carnivors as a case study.\n","tags":["public lecture","GdR EcoStat","statistical ecology"],"title":"Public lecture on statistical ecology","type":"blog"},{"authors":null,"categories":null,"content":"Our new preprint on HMMs for ecology, or how we\u0026rsquo;'\u0026rsquo;ve been using HMMs for ever without even knowing it.\n😎😉 Our new #preprint on #HMM for ecology, or how we\u0026#39;ve been using #HMMs for ever wo even knowing it - all credit goes to @bt_mcclintock and @RolandLangrock for the heavy lifting. Written in @overleaf 😉 @EcoHMM @twitthair1 https://t.co/vSAJF5abH7\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 26020 ","date":1582675200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1582675200,"objectID":"fadf0458ceaaf1fed1e5db37ec594f94","permalink":"https://oliviergimenez.github.io/blog/reviewhmm/","publishdate":"2020-02-26T00:00:00Z","relpermalink":"/blog/reviewhmm/","section":"blog","summary":"Our new preprint on HMMs for ecology, or how we\u0026rsquo;'\u0026rsquo;ve been using HMMs for ever without even knowing it.\n","tags":["HMM","review"],"title":"Review of HMMs in ecology","type":"blog"},{"authors":null,"categories":null,"content":"Landing in Norway to meet w/ Bischof\u0026#39;s group @andyroyle_pwrc @R_nimble and brainstorm about large carnivores and spatially-explicit capture-recapture models #SCR @TouraniMa @PierreDupont47 @Milleret_Cyril the 20 degrees difference w Montpellier is... vivifying 😎🤧 pic.twitter.com/sy0ZUCoae7\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 26, 2020 WildMap happy hour with c/w from top left @eMoqanaki @TouraniMa Daniel Turek @PDupont17 @soumendey_tweet @oaggimenez R. Bischof @Milleret_Cyril Henrik Broseth, yours truly, Jonas Kindberg, @R_nimble De Valpine talking huge-scale monitoring of carnivores, SCR, Nimble, #priceybeer pic.twitter.com/AQqmTeI0Aa\n\u0026mdash; andy 😬 royle (@andyroyle_pwrc) February 26, 2020 😜 We\u0026#39;ve been busy 🐻🐺🚀 https://t.co/4npmtfZLnj pic.twitter.com/WAxqiWrDgu\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 28, 2020 ","date":1582675200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1582675200,"objectID":"23ea63008f851f49160ff2d71dee9162","permalink":"https://oliviergimenez.github.io/blog/norwaywildmap/","publishdate":"2020-02-26T00:00:00Z","relpermalink":"/blog/norwaywildmap/","section":"blog","summary":"Landing in Norway to meet w/ Bischof\u0026#39;s group @andyroyle_pwrc @R_nimble and brainstorm about large carnivores and spatially-explicit capture-recapture models #SCR @TouraniMa @PierreDupont47 @Milleret_Cyril the 20 degrees difference w Montpellier is... vivifying 😎🤧 pic.","tags":["SCR models","large carnivores","Norway"],"title":"Visit to Norway","type":"blog"},{"authors":null,"categories":null,"content":"🇮🇹🇫🇷 @NSantostasi succesfully defended her PhD on modelling population dynamics in presence of hybridization between @SapienzaRoma \u0026amp; @umontpellier #WomenInSTEM #HMM #CaptureRecapture #MatrixModels #IndividualBasedModels #WildlifeManagement pic.twitter.com/grqiuP9xMG\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 21020 ","date":1582243200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1582243200,"objectID":"af872f2f3fec24d51ac936d8a8d8aca7","permalink":"https://oliviergimenez.github.io/blog/ninasdefense/","publishdate":"2020-02-21T00:00:00Z","relpermalink":"/blog/ninasdefense/","section":"blog","summary":"🇮🇹🇫🇷 @NSantostasi succesfully defended her PhD on modelling population dynamics in presence of hybridization between @SapienzaRoma \u0026amp; @umontpellier #WomenInSTEM #HMM #CaptureRecapture #MatrixModels #IndividualBasedModels #WildlifeManagement pic.twitter.com/grqiuP9xMG\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 21020 ","tags":["HMM","wolf demography","hybrids"],"title":"Nina's PhD defense on modelling population dynamics in presence of hybridization","type":"blog"},{"authors":null,"categories":null,"content":"#AcademicTwitter Thought I\u0026#39;d share this excellent thread in English explaining the devastating consequences of the reform of academia (universities and research institutes) and the pension reform in France - kuddos to @MHoussayH for this huge and worth effort 👏 https://t.co/q8TlnOIiKX\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 18020 ","date":1581984000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1581984000,"objectID":"91d1f326ea29724ac0d13723e20b6c31","permalink":"https://oliviergimenez.github.io/blog/reform/","publishdate":"2020-02-18T00:00:00Z","relpermalink":"/blog/reform/","section":"blog","summary":"#AcademicTwitter Thought I\u0026#39;d share this excellent thread in English explaining the devastating consequences of the reform of academia (universities and research institutes) and the pension reform in France - kuddos to @MHoussayH for this huge and worth effort 👏 https://t.","tags":["reform of academia"],"title":"Reform of academia","type":"blog"},{"authors":null,"categories":null,"content":"We had to dig out some #rstats #jags code to fit a structural equation capture-recapture model we developed w/ @SCubi25 in a 8-year old @ESAEcology paper (!), and it runs like a charm 🥳🍾#reproducibility ➡️ https://t.co/Te6BKnxgMF pic.twitter.com/78tWbOhqZD\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 6, 2020 ","date":1580947200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1580947200,"objectID":"2ef67a8e8de92e707d9a8b694855669b","permalink":"https://oliviergimenez.github.io/blog/semjags/","publishdate":"2020-02-06T00:00:00Z","relpermalink":"/blog/semjags/","section":"blog","summary":"We had to dig out some #rstats #jags code to fit a structural equation capture-recapture model we developed w/ @SCubi25 in a 8-year old @ESAEcology paper (!), and it runs like a charm 🥳🍾#reproducibility ➡️ https://t.","tags":["capture-recapture","SEM","rstats"],"title":"Bayesian SEM capture-recapture models","type":"blog"},{"authors":null,"categories":null,"content":"🇬🇷🐬📈@NSantostasi shows in her new paper https://t.co/EZIvJFZbV1 that Common dolphins in the Gulf of Corinth are Critically Endangered (w/ a stochastic PVA) #WomenInSTEM @IUCNRedList #FreeAccess 🇫🇷🇮🇹 pic.twitter.com/eC0plbVgeW\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 2020 ","date":1580601600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1580601600,"objectID":"9a4d6cd57ef01206c0c1de4774ad49b6","permalink":"https://oliviergimenez.github.io/blog/ninapva/","publishdate":"2020-02-02T00:00:00Z","relpermalink":"/blog/ninapva/","section":"blog","summary":"🇬🇷🐬📈@NSantostasi shows in her new paper https://t.co/EZIvJFZbV1 that Common dolphins in the Gulf of Corinth are Critically Endangered (w/ a stochastic PVA) #WomenInSTEM @IUCNRedList #FreeAccess 🇫🇷🇮🇹 pic.twitter.com/eC0plbVgeW\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) February 2020 ","tags":["PVA","Gulf of Corinth","rstats"],"title":"Dolphin PVA","type":"blog"},{"authors":null,"categories":null,"content":"Hello, everyone, this our #FirstTweet We are a group of @CNRS_OccitaniE and @EPHE_fr researchers working at the interface of #ecology, #modelling and #SocialSciences. HAIR is for Human-Animal InteRactions https://t.co/cCwVEmPaCE (new website to come soon) pic.twitter.com/zZX23MQ8L7\n\u0026mdash; twitt\u0026#39;hair (@twitthair1) January 28, 2020 ","date":1580169600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1580169600,"objectID":"6ddedb96b34f0ec37de1a82d0c31fd5d","permalink":"https://oliviergimenez.github.io/blog/team/","publishdate":"2020-01-28T00:00:00Z","relpermalink":"/blog/team/","section":"blog","summary":"Hello, everyone, this our #FirstTweet We are a group of @CNRS_OccitaniE and @EPHE_fr researchers working at the interface of #ecology, #modelling and #SocialSciences. HAIR is for Human-Animal InteRactions https://t.co/cCwVEmPaCE (new website to come soon) pic.","tags":["human-wildlife interactions","research team"],"title":"Our new team is on !","type":"blog"},{"authors":null,"categories":null,"content":"New @biorxivpreprint #preprint led by Sarah Bauduin 🐺💻 From ind behavior and pack dynamics to pop responses: An #IBM to model #wolf social life cycle https://t.co/ea31RvDMoE w/ contributions by @NSantostasi \u0026amp; @oksanagrente #rstats code ➡️ https://t.co/r0hSZYyTpe #WomenInSTEM pic.twitter.com/i8e0C9Ow8w\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) January 26, 2020 ","date":1579996800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1579996800,"objectID":"cf55a29e4a31bf3e30b322fecfd82d79","permalink":"https://oliviergimenez.github.io/blog/ibmwolf/","publishdate":"2020-01-26T00:00:00Z","relpermalink":"/blog/ibmwolf/","section":"blog","summary":"New @biorxivpreprint #preprint led by Sarah Bauduin 🐺💻 From ind behavior and pack dynamics to pop responses: An #IBM to model #wolf social life cycle https://t.co/ea31RvDMoE w/ contributions by @NSantostasi \u0026amp; @oksanagrente #rstats code ➡️ https://t.","tags":["IBM","wolf population dynamics","rstats"],"title":"Wolf population dynamics and individual-based models","type":"blog"},{"authors":null,"categories":null,"content":"Réunion animation du réseau loup/lynx pour l\u0026#39;#hérault #département34 @Occitanie à @villelodeve @OFBiodiversite - un effort de terrain conséquent https://t.co/d6kYRBuilD et précieux https://t.co/Vvem1it4S3 pour la recherche en écologie #SciencesCitoyennes pic.twitter.com/8TbvvotPp9\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) January 20, 2020 ","date":1579478400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1579478400,"objectID":"6a2a36be312a7b5c87ead8a34f254027","permalink":"https://oliviergimenez.github.io/blog/reseauloup/","publishdate":"2020-01-20T00:00:00Z","relpermalink":"/blog/reseauloup/","section":"blog","summary":"Réunion animation du réseau loup/lynx pour l\u0026#39;#hérault #département34 @Occitanie à @villelodeve @OFBiodiversite - un effort de terrain conséquent https://t.co/d6kYRBuilD et précieux https://t.co/Vvem1it4S3 pour la recherche en écologie #SciencesCitoyennes pic.twitter.com/8TbvvotPp9\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) January 20, 2020 ","tags":["loup","gestion des populations"],"title":"Réunion animation du réseau loup/lynx","type":"blog"},{"authors":null,"categories":null,"content":"We had the visit of Eivind F. Kleiven who\u0026#39;s working on small rodents in 🇳🇴 \u0026amp; the fascinating lemmings (awesome paper ➡️ https://t.co/9mPlFNbfd4) - another stimulating collaboration w/ N. Yoccoz, R. Ims \u0026amp; @FredBarraquand \u0026quot;we came to Norway for polar bears, we\u0026#39;ll stay for lemmings\u0026quot; pic.twitter.com/YxfWryID1S\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) January 17, 2020 ","date":1579219200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1579219200,"objectID":"547c2a7f0360600d3bd3f820344d48a8","permalink":"https://oliviergimenez.github.io/blog/eivindvisit/","publishdate":"2020-01-17T00:00:00Z","relpermalink":"/blog/eivindvisit/","section":"blog","summary":"We had the visit of Eivind F. Kleiven who\u0026#39;s working on small rodents in 🇳🇴 \u0026amp; the fascinating lemmings (awesome paper ➡️ https://t.co/9mPlFNbfd4) - another stimulating collaboration w/ N. Yoccoz, R.","tags":["lemmings","animal demography","lemmings","multispecies occupancy"],"title":"Eivind F. Kleiven's visit and lemmings pop dynamics","type":"blog"},{"authors":null,"categories":null,"content":"☘️🚜🍇 New #preprint led by @kazakou_elena in which we assess the effect of management practices on weed demography w/ hidden Markov models #HMM and #RandomEffects #WomenInSTEM #WomenInScience #OpenBUGS pic.twitter.com/xH5mQiMzd8\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) January 9, 2020 ","date":1578528000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1578528000,"objectID":"1d90c9ad6448184a0083f00a02086e33","permalink":"https://oliviergimenez.github.io/blog/elenispaper/","publishdate":"2020-01-09T00:00:00Z","relpermalink":"/blog/elenispaper/","section":"blog","summary":"☘️🚜🍇 New #preprint led by @kazakou_elena in which we assess the effect of management practices on weed demography w/ hidden Markov models #HMM and #RandomEffects #WomenInSTEM #WomenInScience #OpenBUGS pic.twitter.com/xH5mQiMzd8\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) January 9, 2020 ","tags":["HMM","plant demography","weed ecology"],"title":"Weed demography and hidden Markov models","type":"blog"},{"authors":null,"categories":null,"content":"Happy 2020 #rstats! To celebrate, my modest first #TidyTuesday submission for wk 2019-52. Spatio-temporal trends in 🐺 presence in 🇫🇷 w/ data from @oncfs @OFBiodiversite (bugged dynamic map below)#ggplot2 #tidyverse #dataviz #sf @thomas_mock Code: https://t.co/eKkehXlTJH pic.twitter.com/hHwkOmSamR\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) January 2, 2020 ","date":1577923200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1577923200,"objectID":"8f7532d5d369247862409697c65cdc8a","permalink":"https://oliviergimenez.github.io/blog/tidytuesday/","publishdate":"2020-01-02T00:00:00Z","relpermalink":"/blog/tidytuesday/","section":"blog","summary":"Happy 2020 #rstats! To celebrate, my modest first #TidyTuesday submission for wk 2019-52. Spatio-temporal trends in 🐺 presence in 🇫🇷 w/ data from @oncfs @OFBiodiversite (bugged dynamic map below)#ggplot2 #tidyverse #dataviz #sf @thomas_mock Code: https://t.","tags":["tidytuesday","rstats"],"title":"Tidytuesday","type":"blog"},{"authors":null,"categories":null,"content":"","date":1577836800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1577836800,"objectID":"8576ec274c98b3831668a172fa632d80","permalink":"https://oliviergimenez.github.io/about/","publishdate":"2020-01-01T00:00:00Z","relpermalink":"/about/","section":"","summary":"About","tags":null,"title":"About","type":"widget_page"},{"authors":null,"categories":null,"content":"","date":1577836800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1577836800,"objectID":"8e7bc052bdfc6746ea2bb6595e8093eb","permalink":"https://oliviergimenez.github.io/home/","publishdate":"2020-01-01T00:00:00Z","relpermalink":"/home/","section":"","summary":"About","tags":null,"title":"About","type":"widget_page"},{"authors":null,"categories":null,"content":"","date":1577836800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1577836800,"objectID":"3960dd3bdc6f629fb800d1d2aaa7224f","permalink":"https://oliviergimenez.github.io/resume/","publishdate":"2020-01-01T00:00:00Z","relpermalink":"/resume/","section":"","summary":"Resume","tags":null,"title":"Resume","type":"widget_page"},{"authors":null,"categories":null,"content":"Our team has had a fantastic 2019 year! Below is a thread w/ some highlights 🥳🍻📈 @LMarescot @NSantostasi @MaudQueroue @ValentinLauret @oksanagrente @GueryLorelei @roques_seb @SCubi25 @LouvrierJulie https://t.co/FaP9o0bOD4 pic.twitter.com/TuAVaAYkIT\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 29, 2019 2/7. We published papers on the management and #conservation of #wildlife wildlife resorting to #StatisticalEcology and human \u0026amp; social sciences https://t.co/Xlh6reUUQc @EcographyJourna @MethodsEcolEvol @Ecol_Evol @ConBiology @AnimalConserv @JAppliedEcology @biorxivpreprint pic.twitter.com/UbGFGJmmoQ\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 29, 2019 4/7. We engaged w/ stakeholders to try and have an impact on wildlife management. Several projects are on tracks in that direction https://t.co/R4UcwpDxVN pic.twitter.com/Y44sEcyhYb\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 29, 2019 6/7. We are on cloud 9 and lucky to have secured permanent positions in the team #SarahCubaynes (left) @EPHE_fr, #SarahBauduin (middle) @oncfs and #GillesMaurer (right) @zoobeauval pic.twitter.com/T7GRF9LFet\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 29, 2019 Happy New Year! pic.twitter.com/ZMMl4NX7ZK\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 29, 2019 ","date":1577577600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1577577600,"objectID":"9f92a15e23975e9c4688052159f79959","permalink":"https://oliviergimenez.github.io/blog/2019retrospective/","publishdate":"2019-12-29T00:00:00Z","relpermalink":"/blog/2019retrospective/","section":"blog","summary":"Our team has had a fantastic 2019 year! Below is a thread w/ some highlights 🥳🍻📈 @LMarescot @NSantostasi @MaudQueroue @ValentinLauret @oksanagrente @GueryLorelei @roques_seb @SCubi25 @LouvrierJulie https://t.co/FaP9o0bOD4 pic.twitter.com/TuAVaAYkIT\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 29, 2019 ","tags":["retrospective","boulet team"],"title":"Looking back at 2019","type":"blog"},{"authors":null,"categories":null,"content":"Procrastination... Je me suis \u0026quot;amusé\u0026quot; à reproduire cette géniale figure de @coulmont sur le jour des décès en 🇫🇷 en fonction de l\u0026#39;âge et du temps avec #rstats et le #tidyverse https://t.co/W8QNBX5WlT #ggplot2 #dplyr #lubridate https://t.co/wOthJOBzBd\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 23, 2019 ","date":1577059200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1577059200,"objectID":"7a3bb5d061fecf335036b1d7f13d77a5","permalink":"https://oliviergimenez.github.io/blog/procrastination_coulmont/","publishdate":"2019-12-23T00:00:00Z","relpermalink":"/blog/procrastination_coulmont/","section":"blog","summary":"Procrastination... Je me suis \u0026quot;amusé\u0026quot; à reproduire cette géniale figure de @coulmont sur le jour des décès en 🇫🇷 en fonction de l\u0026#39;âge et du temps avec #rstats et le #tidyverse https://t.","tags":["tidyverse","demographie"],"title":"Mortalité en France et tidyverse","type":"blog"},{"authors":null,"categories":null,"content":"We\u0026#39;ve been thinking a lot about #predictions in ecology lately. Here\u0026#39;s our modest crack at it https://t.co/NU7ROWLFWF led by @LouvrierJulie w/ a mechanistic-statistical #SpeciesDistributionModel to #forecast 🇫🇷🐺 expansion #ImperfectDetection #WomenInSTEM #ODEs #Bayes pic.twitter.com/D1n38edpDd\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 23, 2019 ","date":1577059200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1577059200,"objectID":"d4c711e4e0e3a35606c82f5584acb242","permalink":"https://oliviergimenez.github.io/blog/preprintmecastat/","publishdate":"2019-12-23T00:00:00Z","relpermalink":"/blog/preprintmecastat/","section":"blog","summary":"We\u0026#39;ve been thinking a lot about #predictions in ecology lately. Here\u0026#39;s our modest crack at it https://t.co/NU7ROWLFWF led by @LouvrierJulie w/ a mechanistic-statistical #SpeciesDistributionModel to #forecast 🇫🇷🐺 expansion #ImperfectDetection #WomenInSTEM #ODEs #Bayes pic.","tags":["SDM","wolf","management","prediction","forecasting"],"title":"New preprint by Julie Louvrier on forecasting species distribution","type":"blog"},{"authors":null,"categories":null,"content":"Playing around w/ awesome @R_nimble to fit multievent #HMM capture-recapture models to data - Code and data here https://t.co/2oHvsplahv Comments welcome 🤗 pic.twitter.com/tA89r80YYV\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 2019 ","date":1576800000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1576800000,"objectID":"1ee1fff67419804caa211258bd72966d","permalink":"https://oliviergimenez.github.io/blog/nimblehmm/","publishdate":"2019-12-20T00:00:00Z","relpermalink":"/blog/nimblehmm/","section":"blog","summary":"Playing around w/ awesome @R_nimble to fit multievent #HMM capture-recapture models to data - Code and data here https://t.co/2oHvsplahv Comments welcome 🤗 pic.twitter.com/tA89r80YYV\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 2019 ","tags":["HMM","Nimble","Bayes","MCMC","capture-recapture","multievent"],"title":"Nimble and HMM","type":"blog"},{"authors":null,"categories":null,"content":"Our team will be at the World Marine Mammal Conference in Barcelona @VisitBCN_FR 🐬📊🤩 @ValentinLauret @NSantostasi @GIS_3M #WMMC19 @wmmc2019 #MarineMammals - check out the who, what, when and where below pic.twitter.com/eaG36U4e9g\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) December 3, 2019 ","date":1575331200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1575331200,"objectID":"291bf36b9a93361eb43b6952f69a4f33","permalink":"https://oliviergimenez.github.io/blog/wmmc/","publishdate":"2019-12-03T00:00:00Z","relpermalink":"/blog/wmmc/","section":"blog","summary":"Our team will be at the World Marine Mammal Conference in Barcelona @VisitBCN_FR 🐬📊🤩 @ValentinLauret @NSantostasi @GIS_3M #WMMC19 @wmmc2019 #MarineMammals - check out the who, what, when and where below pic.","tags":["marine mammal","conservation biology"],"title":"World Marine Mammal Conference","type":"blog"},{"authors":null,"categories":null,"content":"This week we host our first winter school on «Reproducible Research in Numerical Ecology » co-organised by #GDR_Ecostat \u0026amp; #CESAB @FRBiodiv. Many thanks to our incredible list of speakers : Nicolas Casajus, Stéphane Dray, @GueryLorelei, @oaggimenez, @FGuilhaumon \u0026amp; @NinaSchiett !! pic.twitter.com/lRKlVXtYym\n\u0026mdash; Nicolas Mouquet (@NicolasMouquet) December 2,2019 ","date":1575244800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1575244800,"objectID":"e5674a0bbd3a91f9ae7e61346da11136","permalink":"https://oliviergimenez.github.io/blog/cesabworkshop/","publishdate":"2019-12-02T00:00:00Z","relpermalink":"/blog/cesabworkshop/","section":"blog","summary":"This week we host our first winter school on «Reproducible Research in Numerical Ecology » co-organised by #GDR_Ecostat \u0026amp; #CESAB @FRBiodiv. Many thanks to our incredible list of speakers : Nicolas Casajus, Stéphane Dray, @GueryLorelei, @oaggimenez, @FGuilhaumon \u0026amp; @NinaSchiett !","tags":["reproducible science","cesab","gdr ecostat","statistical ecology","tidyverse"],"title":"Workshop on reproducible science","type":"blog"},{"authors":null,"categories":null,"content":"New @biorxivpreprint #preprint in which @ValentinLauret combines aerial \u0026amp; boat surveys in #SDM for bottlenose dolphins w/ #occupancy models (single and repeated visits) - w/ H. Labach \u0026amp; M. Authier https://t.co/t2IJMTbUf8 pic.twitter.com/F2ftXu0DtK\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) November 21 2019 ","date":1574294400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1574294400,"objectID":"7004475982c7fd02ddc400458b945d20","permalink":"https://oliviergimenez.github.io/blog/preprintval/","publishdate":"2019-11-21T00:00:00Z","relpermalink":"/blog/preprintval/","section":"blog","summary":"New @biorxivpreprint #preprint in which @ValentinLauret combines aerial \u0026amp; boat surveys in #SDM for bottlenose dolphins w/ #occupancy models (single and repeated visits) - w/ H. Labach \u0026amp; M. Authier https://t.","tags":["SDM","tursiops","conservation","bottlenose dolphin","data integration"],"title":"New preprint by Valentin Lauret on integrated SDMs","type":"blog"},{"authors":null,"categories":null,"content":"Awesome Nimble @R_nimble workshop given by Daniel Turek from @WilliamsCollege #Bayes #StatisticalEcology at @CNRS_OccitaniE @INEE_CNRS @umontpellier w/ @oksanagrente @ValentinLauret @MaudQueroue @roques_seb @GSouchay @DavidVallecill_ @LoBacon1 pic.twitter.com/7zjQWKeF9P\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) October 23, 2019 ","date":1571788800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1571788800,"objectID":"fddad995a465f503e020e51d76a7869c","permalink":"https://oliviergimenez.github.io/blog/nimbleworkshop/","publishdate":"2019-10-23T00:00:00Z","relpermalink":"/blog/nimbleworkshop/","section":"blog","summary":"Awesome Nimble @R_nimble workshop given by Daniel Turek from @WilliamsCollege #Bayes #StatisticalEcology at @CNRS_OccitaniE @INEE_CNRS @umontpellier w/ @oksanagrente @ValentinLauret @MaudQueroue @roques_seb @GSouchay @DavidVallecill_ @LoBacon1 pic.twitter.com/7zjQWKeF9P\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) October 23, 2019 ","tags":["workshop","Nimble","Bayesian statistics"],"title":"Nimble workshop with Daniel Turek","type":"blog"},{"authors":null,"categories":null,"content":"Formation #CESAB / GDR EcoStat: Bonnes pratiques pour une recherche reproductible en écologie numérique, 2-6/12/2019, Montpellier. Par Nicolas Casajus, Stéphane Dray, @GueryLorelei, @oaggimenez @FGuilhaumon \u0026amp; @NinaSchiett\n➡https://t.co/P0HVLMEf7f@FRBiodiv @INEE_CNRS @Ifremer_fr pic.twitter.com/F9jCllLw6X\n\u0026mdash; Nicolas Mouquet (@NicolasMouquet) October 15, 2019 ","date":1571097600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1571097600,"objectID":"aac0b28f18a475f476f48aaf82ae4057","permalink":"https://oliviergimenez.github.io/blog/cesab/","publishdate":"2019-10-15T00:00:00Z","relpermalink":"/blog/cesab/","section":"blog","summary":"Formation #CESAB / GDR EcoStat: Bonnes pratiques pour une recherche reproductible en écologie numérique, 2-6/12/2019, Montpellier. Par Nicolas Casajus, Stéphane Dray, @GueryLorelei, @oaggimenez @FGuilhaumon \u0026amp; @NinaSchiett\n➡https://t.co/P0HVLMEf7f@FRBiodiv @INEE_CNRS @Ifremer_fr pic.twitter.com/F9jCllLw6X","tags":["workshop","ecostat","reproducible science"],"title":"Workshop on reproducible science","type":"blog"},{"authors":null,"categories":null,"content":"A thread on my review of the capture-recapture literature for the conference Wildlife Research and Conservation 2019 https://t.co/dqz8kbLdym in Berlin 🇩🇪. Slides, analyses, data \u0026amp; #rstats codes available from https://t.co/s9sFbkZG4k 1/23 pic.twitter.com/crGkVqYzov\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) September 28, /a\u003e ","date":1569628800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1569628800,"objectID":"903e1d7c8e43d1413172ab058242de98","permalink":"https://oliviergimenez.github.io/blog/crreview/","publishdate":"2019-09-28T00:00:00Z","relpermalink":"/blog/crreview/","section":"blog","summary":"A thread on my review of the capture-recapture literature for the conference Wildlife Research and Conservation 2019 https://t.co/dqz8kbLdym in Berlin 🇩🇪. Slides, analyses, data \u0026amp; #rstats codes available from https://t.co/s9sFbkZG4k 1/23 pic.","tags":["capture-recapture","systematic review"],"title":"A review on the last decade of research on capture-recapture","type":"blog"},{"authors":null,"categories":null,"content":"🥳🍾 Our paper providing lynx density estimates for 🇫🇷 Jura and Vosges mountains is out in @Ecol_Evol https://t.co/74rA1iUaQU @oncfs @CROC_Estelle @ZiFridolin #CameraTrap #SCR #spatial #capturerecapture w/ @oSCR_package #rstats #openaccess https://t.co/s8ooMzNDVI\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) September 28,2019 ","date":1569628800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1569628800,"objectID":"f7c073bb8a304365347ae652f9b595ee","permalink":"https://oliviergimenez.github.io/blog/paper_lynxdensities/","publishdate":"2019-09-28T00:00:00Z","relpermalink":"/blog/paper_lynxdensities/","section":"blog","summary":"🥳🍾 Our paper providing lynx density estimates for 🇫🇷 Jura and Vosges mountains is out in @Ecol_Evol https://t.co/74rA1iUaQU @oncfs @CROC_Estelle @ZiFridolin #CameraTrap #SCR #spatial #capturerecapture w/ @oSCR_package #rstats #openaccess https://t.co/s8ooMzNDVI","tags":["lynx","SCR models","population densities"],"title":"New paper on lynx densities in France","type":"blog"},{"authors":null,"categories":null,"content":"Découverte par les acteurs de l\u0026#39;interface logiciel pour évaluer l\u0026#39;impact d\u0026#39;aménagements sur la viabilité du lynx - 3ème atelier co-construction projet \u0026quot;Eviter, réduire \u0026amp; compenser la mortalité du Lynx par collision avec les véhicules de transport\u0026quot; https://t.co/uKY5ylOli2 #ITTECOP pic.twitter.com/0MWDBraOa3\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) September 25,2019 ","date":1569369600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1569369600,"objectID":"5c3f51e3014d2eff1a383a17d646b537","permalink":"https://oliviergimenez.github.io/blog/atelier3erclynx/","publishdate":"2019-09-25T00:00:00Z","relpermalink":"/blog/atelier3erclynx/","section":"blog","summary":"Découverte par les acteurs de l\u0026#39;interface logiciel pour évaluer l\u0026#39;impact d\u0026#39;aménagements sur la viabilité du lynx - 3ème atelier co-construction projet \u0026quot;Eviter, réduire \u0026amp; compenser la mortalité du Lynx par collision avec les véhicules de transport\u0026quot; https://t.","tags":["lynx","co-construction"],"title":"Co-construction d'une interface logiciel","type":"blog"},{"authors":null,"categories":null,"content":"New #preprint by @MGamelon on \u0026quot;Efficient use of harvest data: An integrated population model for exploited animal populations\u0026quot; https://t.co/s2nVkphPTe in which we combine data on dead and alive animals to infer population dynamics of the fascinating wild boar pic.twitter.com/MuJNXWjYZ5\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) September 20 2019 ","date":1568937600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1568937600,"objectID":"6d758f1ee39e00222a7be373444fef79","permalink":"https://oliviergimenez.github.io/blog/paper_ipmboar/","publishdate":"2019-09-20T00:00:00Z","relpermalink":"/blog/paper_ipmboar/","section":"blog","summary":"New #preprint by @MGamelon on \u0026quot;Efficient use of harvest data: An integrated population model for exploited animal populations\u0026quot; https://t.co/s2nVkphPTe in which we combine data on dead and alive animals to infer population dynamics of the fascinating wild boar pic.","tags":["IPM","wild boar"],"title":"New preprint: Efficient use of harvest data: An integrated population model for exploited animal populations","type":"blog"},{"authors":null,"categories":null,"content":"Very happy that our article on fitting species distribution models via combined likelihood to presence-only and occupancy data with lasso penalties and accounting for spatial dependence has been published in MEE! Go team! @LouvrierJulie @oaggimenez 🇦🇺🇫🇷🧀https://t.co/apVUbNb7Fi\n\u0026mdash; Ian Renner (@MoonbeamLevels) September 1819 ","date":1568764800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1568764800,"objectID":"8ee45ef39d4ccdd94e5c442a676e8782","permalink":"https://oliviergimenez.github.io/blog/sdmpaper/","publishdate":"2019-09-18T00:00:00Z","relpermalink":"/blog/sdmpaper/","section":"blog","summary":"Very happy that our article on fitting species distribution models via combined likelihood to presence-only and occupancy data with lasso penalties and accounting for spatial dependence has been published in MEE!","tags":["species distribution models","data combination"],"title":"New paper on SDMs and data combination","type":"blog"},{"authors":null,"categories":null,"content":"Stimulating 🇨🇭🇫🇷 meeting to discuss transboundary monitoring and research on 🐱🐺 in the Jura ⛰️ #KORA @oncfs (@INEE_CNRS) C. Duchamp, @ZiFridolin, U. Breitenmoser, C. Breitenmoser-Würsten, and N. Drouet-Hoguet not on the pix pic.twitter.com/WInB8uh1Vh\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) September 13, 20/a\u003e ","date":1568332800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1568332800,"objectID":"a309d6d210cd4dd04ea302a69eaa6de7","permalink":"https://oliviergimenez.github.io/blog/koraoncfsmeeting/","publishdate":"2019-09-13T00:00:00Z","relpermalink":"/blog/koraoncfsmeeting/","section":"blog","summary":"Stimulating 🇨🇭🇫🇷 meeting to discuss transboundary monitoring and research on 🐱🐺 in the Jura ⛰️ #KORA @oncfs (@INEE_CNRS) C. Duchamp, @ZiFridolin, U. Breitenmoser, C. Breitenmoser-Würsten, and N. Drouet-Hoguet not on the pix pic.","tags":["lynx","transboundary monitoring","large carnivores"],"title":"Lynx meeting on transboundary monitoring and research","type":"blog"},{"authors":null,"categories":null,"content":"🥳 It is a pleasure to welcome Daniel Turek @WilliamsCollege for a 6-month sabbatical here in @montpellier_ Stay tuned for more about #Bayes #StatisticalEcology w/ @R_nimble 🤩 Check out Daniel\u0026#39;s website ▶️ https://t.co/HYyMMrN6Nb pic.twitter.com/vK7q6bYzPU\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) September 919 ","date":1567987200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1567987200,"objectID":"491d8d4ee1463010d2b0f7529da82d80","permalink":"https://oliviergimenez.github.io/blog/daniel_turek/","publishdate":"2019-09-09T00:00:00Z","relpermalink":"/blog/daniel_turek/","section":"blog","summary":"🥳 It is a pleasure to welcome Daniel Turek @WilliamsCollege for a 6-month sabbatical here in @montpellier_ Stay tuned for more about #Bayes #StatisticalEcology w/ @R_nimble 🤩 Check out Daniel\u0026#39;s website ▶️ https://t.","tags":["sabbatical","Nimble","Bayes"],"title":"Daniel Turek is with us for 6 months!","type":"blog"},{"authors":null,"categories":null,"content":"New awesome paper by Sarah Bauduin https://t.co/Axu1kkkdSO in which she and her co-authors introduce #NetLogoR to build individual-based models w/ R #rstats 🥳🤩 Stay tuned as @oksanagrente and @NSantostasi are using #NetLogoR to study pop management and hybridization w/ wolf 🐺 https://t.co/fddUqENrAw\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) August 13, 2019 ","date":1565654400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1565654400,"objectID":"dc3de34c7b7571dc93a129bbe97be8c7","permalink":"https://oliviergimenez.github.io/blog/paper_netlogor/","publishdate":"2019-08-13T00:00:00Z","relpermalink":"/blog/paper_netlogor/","section":"blog","summary":"New awesome paper by Sarah Bauduin https://t.co/Axu1kkkdSO in which she and her co-authors introduce #NetLogoR to build individual-based models w/ R #rstats 🥳🤩 Stay tuned as @oksanagrente and @NSantostasi are using #NetLogoR to study pop management and hybridization w/ wolf 🐺 https://t.","tags":["R package","individual-based models"],"title":"Awesome paper introducing NetLogoR to build individual-based models with R","type":"blog"},{"authors":null,"categories":null,"content":"New #preprint w/ @GIS_3M on Distribution and abundance of bottlenose dolphin (Tursiops truncatus) over the French Mediterranean continental shelf https://t.co/eNfgLCT4k4 - Soooo impressed by the huge collective field effort to collect photo-id on this iconic species! pic.twitter.com/gRqpzkg6UK\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) August 3, 2019 ","date":1564790400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1564790400,"objectID":"fc50de92b800b14c7d41b25696afef72","permalink":"https://oliviergimenez.github.io/blog/paper_hlabach/","publishdate":"2019-08-03T00:00:00Z","relpermalink":"/blog/paper_hlabach/","section":"blog","summary":"New #preprint w/ @GIS_3M on Distribution and abundance of bottlenose dolphin (Tursiops truncatus) over the French Mediterranean continental shelf https://t.co/eNfgLCT4k4 - Soooo impressed by the huge collective field effort to collect photo-id on this iconic species!","tags":["dolphins","tursiops","capture-recapture","abundance"],"title":"New preprint: Distribution and abundance of bottlenose dolphin (Tursiops truncatus) over the French Mediterranean continental shelf","type":"blog"},{"authors":null,"categories":null,"content":"Le projet porté par Gilles Maurer dans lequel nous étudions les conditions de la #coexistence entre humains et #éléphant d\u0026#39;Asie mis à l\u0026#39;honneur par @IsiteMUSE @umontpellier - collaborations entre @zoobeauval @INEE_CNRS @CNRS_OccitaniE @INSHS_CNRS pic.twitter.com/VvAYsYWDSc\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 19 2019 ","date":1563494400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1563494400,"objectID":"09bac254ab408729833065137b8e1612","permalink":"https://oliviergimenez.github.io/blog/peace/","publishdate":"2019-07-19T00:00:00Z","relpermalink":"/blog/peace/","section":"blog","summary":"Le projet porté par Gilles Maurer dans lequel nous étudions les conditions de la #coexistence entre humains et #éléphant d\u0026#39;Asie mis à l\u0026#39;honneur par @IsiteMUSE @umontpellier - collaborations entre @zoobeauval @INEE_CNRS @CNRS_OccitaniE @INSHS_CNRS pic.","tags":["éléphant","Muse","PEACE"],"title":"Gilles et le projet Muse sur la conservation de l'éléphant d'Asie","type":"blog"},{"authors":null,"categories":null,"content":"Excellent article \u0026quot;Doit-on (vraiment) crier au loup\u0026quot; par @CJosset @LEXPRESS - @oncfs @INEE_CNRS @CNRS_OccitaniE @IsiteMUSE @umontpellier https://t.co/PpmvNnpM9Y\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 15, 2019 ","date":1563148800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1563148800,"objectID":"d700c33a08d916d29050ac54e84ea235","permalink":"https://oliviergimenez.github.io/blog/article_loup_lexpress/","publishdate":"2019-07-15T00:00:00Z","relpermalink":"/blog/article_loup_lexpress/","section":"blog","summary":"Excellent article \u0026quot;Doit-on (vraiment) crier au loup\u0026quot; par @CJosset @LEXPRESS - @oncfs @INEE_CNRS @CNRS_OccitaniE @IsiteMUSE @umontpellier https://t.co/PpmvNnpM9Y\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) July 15, 2019 ","tags":["loup","presse"],"title":"Article sur le loup dans L'Express","type":"blog"},{"authors":null,"categories":null,"content":"🥳🍾 Awesome work by AS Bonnet Lebrun, A Karamanlidis from @Arcturosgr M de Gabriel Hernando and I Renner @MoonbeamLevels in which we use Poisson point process models to identify priority conservation areas for brown bears in Greece 🇬🇷🐻📈 https://t.co/rDV7h9Zb7C pic.twitter.com/UmgmynE3Jd\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 5 juillet19 ","date":1562284800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1562284800,"objectID":"c6744c1dd6fb6b98f996e26e7c119715","permalink":"https://oliviergimenez.github.io/blog/paperanneso/","publishdate":"2019-07-05T00:00:00Z","relpermalink":"/blog/paperanneso/","section":"blog","summary":"🥳🍾 Awesome work by AS Bonnet Lebrun, A Karamanlidis from @Arcturosgr M de Gabriel Hernando and I Renner @MoonbeamLevels in which we use Poisson point process models to identify priority conservation areas for brown bears in Greece 🇬🇷🐻📈 https://t.","tags":["point process","species distribution model","brown bear","large carnivores","Greece"],"title":"New paper: Identifying priority conservation areas for a recovering brown bear population in Greece using citizen science data","type":"blog"},{"authors":null,"categories":null,"content":"New paper by @TamarLok 🤩👏 using integrated modelling of 8 years of resighting and count data and accouting for heterogeneity to show the key importance of the Luannan Coast of Bohai Bay (China) as a spring #stopover area for red knots https://t.co/EgOA6yM4Dp #WomenInSTEM pic.twitter.com/05qVXhyXVt\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 28 juin 2019 ","date":1561680000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1561680000,"objectID":"4989c5f0dbaed9dee071d320cea1f317","permalink":"https://oliviergimenez.github.io/blog/papertamar/","publishdate":"2019-06-28T00:00:00Z","relpermalink":"/blog/papertamar/","section":"blog","summary":"New paper by @TamarLok 🤩👏 using integrated modelling of 8 years of resighting and count data and accouting for heterogeneity to show the key importance of the Luannan Coast of Bohai Bay (China) as a spring #stopover area for red knots https://t.","tags":["heterogeneity","demography","capture-recapture"],"title":"New paper: Accounting for heterogeneity when estimating stopover duration, timing and population size of red knots along the Luannan Coast of Bohai Bay, China","type":"blog"},{"authors":null,"categories":null,"content":"Awesome talk by @hgcaceres on \u0026#39;Bottom-up conservation, engaging with the local community to develop better conservation actions\u0026#39; 🤩 @CNRS_OccitaniE @INEE_CNRS @umontpellier @IsiteMUSE pic.twitter.com/dGOGFY9B8x\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 18juin 2019 ","date":1560643200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1560643200,"objectID":"085b23e2409486fd8b0fc2e1ba21a371","permalink":"https://oliviergimenez.github.io/blog/hernanvisit/","publishdate":"2019-06-16T00:00:00Z","relpermalink":"/blog/hernanvisit/","section":"blog","summary":"Awesome talk by @hgcaceres on \u0026#39;Bottom-up conservation, engaging with the local community to develop better conservation actions\u0026#39; 🤩 @CNRS_OccitaniE @INEE_CNRS @umontpellier @IsiteMUSE pic.twitter.com/dGOGFY9B8x\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 18juin 2019 ","tags":["adaptive management","network"],"title":"Seminar by Hernan Caceres on Bottom-up conservation, engaging with the local community to develop better conservation actions","type":"blog"},{"authors":null,"categories":null,"content":"You have gaps in your data time-series? Open-pop SCR models may help you infer abundance, even during the gap years @PDupont17 @joechip90 @oaggimenez @R_nimble 🚀@bxv_ecol @biorxiv_ecology https://t.co/0YjWa1PmIL pic.twitter.com/Aa4cQPAcoP\n\u0026mdash; Cyril Milleret (@Milleret_Cyril) 15 juin 2019 ","date":1560556800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1560556800,"objectID":"7f59b693b0b66c9f0c8f6216b7cce29b","permalink":"https://oliviergimenez.github.io/blog/preprintcyril/","publishdate":"2019-06-15T00:00:00Z","relpermalink":"/blog/preprintcyril/","section":"blog","summary":"You have gaps in your data time-series? Open-pop SCR models may help you infer abundance, even during the gap years @PDupont17 @joechip90 @oaggimenez @R_nimble 🚀@bxv_ecol @biorxiv_ecology https://t.co/0YjWa1PmIL pic.twitter.com/Aa4cQPAcoP\n\u0026mdash; Cyril Milleret (@Milleret_Cyril) 15 juin 2019 ","tags":["scr","large carnivores","spatial capture-recapture"],"title":"New preprint: Estimating abundance with interruptions in data collection using open population spatial capture-recapture models","type":"blog"},{"authors":null,"categories":null,"content":"Wow, the 5th conference of our 🇫🇷 group in #StatisticalEcology is open in the beautiful city of #Avignon @UnivAvignon #GdREcoStat - Check out the program ▶️ https://t.co/1H8z4VRGxI pic.twitter.com/2Fyw1NleJG\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 13 mai 2019 ","date":1557705600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1557705600,"objectID":"f723b2d4f8145203f2c5a6841ec60589","permalink":"https://oliviergimenez.github.io/blog/gdrecostat2019/","publishdate":"2019-05-13T00:00:00Z","relpermalink":"/blog/gdrecostat2019/","section":"blog","summary":"Wow, the 5th conference of our 🇫🇷 group in #StatisticalEcology is open in the beautiful city of #Avignon @UnivAvignon #GdREcoStat - Check out the program ▶️ https://t.co/1H8z4VRGxI pic.twitter.com/2Fyw1NleJG\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 13 mai 2019 ","tags":["statistical ecology","Avignon","GDR"],"title":"5th French conference on Statistical Ecology (thread)","type":"blog"},{"authors":null,"categories":null,"content":"I thought I\u0026#39;d share some #rstats code about local minima in #multistate #capturerecapture models https://t.co/aJE33dBtdO using #Mark #Admb #Jags #Stan (#Nimble \u0026amp; #TMB to come). Largely inspired by Chapter 9 of the \u0026quot;Gentle introduction to Mark\u0026quot; by Cooch \u0026amp; White 😉 pic.twitter.com/NxzG6w0Bzq\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 12 mai 2019 ","date":1557619200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1557619200,"objectID":"e43d5bcac59a60132d8f2347c3135857","permalink":"https://oliviergimenez.github.io/blog/localminima_reboot/","publishdate":"2019-05-12T00:00:00Z","relpermalink":"/blog/localminima_reboot/","section":"blog","summary":"I thought I\u0026#39;d share some #rstats code about local minima in #multistate #capturerecapture models https://t.co/aJE33dBtdO using #Mark #Admb #Jags #Stan (#Nimble \u0026amp; #TMB to come). Largely inspired by Chapter 9 of the \u0026quot;Gentle introduction to Mark\u0026quot; by Cooch \u0026amp; White 😉 pic.","tags":["local minima","capture-recapture"],"title":"Local minima and multistate capture-recapture models","type":"blog"},{"authors":null,"categories":null,"content":"Just wrote a short #tutorial on fitting a system of Ordinary Differential Equations to time-series data using #OpenBUGS in #rstats https://t.co/7luj8XVgP6 - Data and model are from https://t.co/X9CwX7dW83 in which #zombie movies are used for illustration #shaunofthedead pic.twitter.com/MHmiGhH7Ri\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 9 mai 2019 ","date":1557446400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1557446400,"objectID":"f74a377e029288907cce795d8126f188","permalink":"https://oliviergimenez.github.io/blog/fitodedata/","publishdate":"2019-05-10T00:00:00Z","relpermalink":"/blog/fitodedata/","section":"blog","summary":"Just wrote a short #tutorial on fitting a system of Ordinary Differential Equations to time-series data using #OpenBUGS in #rstats https://t.co/7luj8XVgP6 - Data and model are from https://t.co/X9CwX7dW83 in which #zombie movies are used for illustration #shaunofthedead pic.","tags":["ordinary differential equations","OpenBUGS","zombies"],"title":"Tutorial on fitting a system of Ordinary Differential Equations to time-series data using OpenBUGS","type":"blog"},{"authors":null,"categories":null,"content":"A nice post by @ksuryawanshi on our recent paper on transboundary management of wolverines in Scandinavia: \u0026quot;Carnivores without borders: management of transboundary populations when objectives differ\u0026quot; https://t.co/OcAAgwUP4U via @JAppliedEcology\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 9 mai 2019 ","date":1557360000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1557360000,"objectID":"0517a35d02b7663ba51a0fcc42c3fd99","permalink":"https://oliviergimenez.github.io/blog/blogpost_articlevincenzo/","publishdate":"2019-05-09T00:00:00Z","relpermalink":"/blog/blogpost_articlevincenzo/","section":"blog","summary":"A nice post by @ksuryawanshi on our recent paper on transboundary management of wolverines in Scandinavia: \u0026quot;Carnivores without borders: management of transboundary populations when objectives differ\u0026quot; https://t.co/OcAAgwUP4U via @JAppliedEcology\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 9 mai 2019 ","tags":["wolverine","blog post","transboundary management","large carnivores"],"title":"Carnivores without borders: management of transboundary populations when objectives differ","type":"blog"},{"authors":null,"categories":null,"content":"New paper on \u0026#39;How many cubs can a mum nurse? Maternal age and size influence litter size in polar bears\u0026#39; https://t.co/Utu4VZ0Ms3 w/ #DorindaFolio (former Master student 👏🤩) and #SarahCubaynes in control #WomenInSTEM - PDF on request 😉 pic.twitter.com/QABpoQxmHQ\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 1 mai 2019 ","date":1556668800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1556668800,"objectID":"45ec11e2bcd5b271836052a92b57f8c9","permalink":"https://oliviergimenez.github.io/blog/paperdorinda/","publishdate":"2019-05-01T00:00:00Z","relpermalink":"/blog/paperdorinda/","section":"blog","summary":"New paper on \u0026#39;How many cubs can a mum nurse? Maternal age and size influence litter size in polar bears\u0026#39; https://t.co/Utu4VZ0Ms3 w/ #DorindaFolio (former Master student 👏🤩) and #SarahCubaynes in control #WomenInSTEM - PDF on request 😉 pic.","tags":["polar bear","demography","life-history strategies"],"title":"New paper: How many cubs can a mum nurse? Maternal age and size influence litter size in polar bears","type":"blog"},{"authors":null,"categories":null,"content":"New #preprint on \u0026quot;Combining multiple data sources in species distribution models while accounting for spatial dependence and overfitting with penalised likelihood maximisation\u0026quot; https://t.co/7GLRrgpo4D w/ @MoonbeamLevels \u0026amp; @LouvrierJulie #SDM #MachineLearning #PointProcess pic.twitter.com/6xzAlARV9E\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 24 avri 2019 ","date":1556064000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1556064000,"objectID":"abe5152e886e8e0718c000280bcdb663","permalink":"https://oliviergimenez.github.io/blog/preprintian/","publishdate":"2019-04-24T00:00:00Z","relpermalink":"/blog/preprintian/","section":"blog","summary":"New #preprint on \u0026quot;Combining multiple data sources in species distribution models while accounting for spatial dependence and overfitting with penalised likelihood maximisation\u0026quot; https://t.co/7GLRrgpo4D w/ @MoonbeamLevels \u0026amp; @LouvrierJulie #SDM #MachineLearning #PointProcess pic.","tags":["sdm","occupancy","point process","lasso"],"title":"Combining multiple data sources in species distribution models while accounting for spatial dependence and overfitting with penalised likelihood maximisation","type":"blog"},{"authors":null,"categories":null,"content":"Quick and dirty code to fit a multistate single-season #occupancy model in Jags using a #HMM formulation, in case it\u0026#39;s useful https://t.co/vUlrP0jjIU #JAGS #rstats w/ @ArnaudLyet pic.twitter.com/dSr0BlICb6\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 13l 2019 ","date":1555113600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1555113600,"objectID":"c9a2ec26901b631444573711d522d9e0","permalink":"https://oliviergimenez.github.io/blog/coding/","publishdate":"2019-04-13T00:00:00Z","relpermalink":"/blog/coding/","section":"blog","summary":"Quick and dirty code to fit a multistate single-season #occupancy model in Jags using a #HMM formulation, in case it\u0026#39;s useful https://t.co/vUlrP0jjIU #JAGS #rstats w/ @ArnaudLyet pic.twitter.com/dSr0BlICb6\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 13l 2019 ","tags":["occupancy","jags"],"title":"Jags code to fit a multistate single-season occupancy model","type":"blog"},{"authors":null,"categories":null,"content":"New paper on \u0026quot;Inferring animal social networks with imperfect detection\u0026quot; w/ colleagues from 🇦🇷 and #LorenaMansilla our 🇨🇱 PhD student #SocialNetworks #CaptureRecapture #StateSpaceModel #rstats #BayesianStatistics - Download PDF freely here https://t.co/heixxmRdFH pic.twitter.com/dZbRLvFbjT\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 10 avril 2019 ","date":1554854400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1554854400,"objectID":"2e7ed5cb8c90c7fa549a0a83dbdfa00d","permalink":"https://oliviergimenez.github.io/blog/paperog/","publishdate":"2019-04-10T00:00:00Z","relpermalink":"/blog/paperog/","section":"blog","summary":"New paper on \u0026quot;Inferring animal social networks with imperfect detection\u0026quot; w/ colleagues from 🇦🇷 and #LorenaMansilla our 🇨🇱 PhD student #SocialNetworks #CaptureRecapture #StateSpaceModel #rstats #BayesianStatistics - Download PDF freely here https://t.","tags":["social network","dolphin","hmm"],"title":"New paper on Inferring animal social networks with imperfect detection","type":"blog"},{"authors":null,"categories":null,"content":"🐺🔎💩 Un chouette article https://t.co/Qm0fkSX675 dans @maglarecherche sur les travaux de thèse de @LouvrierJulie sur écologie statistique, modèles de distribution d\u0026#39;espèce et grands carnivores - merci @BereniceRbt ! Bel exemple de collaborations @oncfs @INEE_CNRS @umontpellier pic.twitter.com/xkvjC4hbQr\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 7l 2019 ","date":1554595200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1554595200,"objectID":"78c39a02602cdecc901fc48e3c2d9a8f","permalink":"https://oliviergimenez.github.io/blog/commloup/","publishdate":"2019-04-07T00:00:00Z","relpermalink":"/blog/commloup/","section":"blog","summary":"🐺🔎💩 Un chouette article https://t.co/Qm0fkSX675 dans @maglarecherche sur les travaux de thèse de @LouvrierJulie sur écologie statistique, modèles de distribution d\u0026#39;espèce et grands carnivores - merci @BereniceRbt ! Bel exemple de collaborations @oncfs @INEE_CNRS @umontpellier pic.","tags":["loup","occupancy"],"title":"Un article dans La Recherche sur nos travaux","type":"blog"},{"authors":null,"categories":null,"content":"How many lynx are there? \u0026quot;Spatial density estimates of Eurasian lynx (Lynx lynx) in the 🇫🇷 Jura and Vosges 🏔️\u0026quot; https://t.co/roLlOsFxw7 @oncfs @CROC_Estelle @ZiFridolin w/ awesome #oSCR @rstats 📦 @chrissuthy @andyroyle_pwrc @dwlinden @CNRS_OccitaniE @INEE_CNRS pic.twitter.com/AsUGJVhmKx\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 5 avril 2019 ","date":1554422400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1554422400,"objectID":"872f7082c34bdd54ccaf34412c5c5e26","permalink":"https://oliviergimenez.github.io/blog/preprintog/","publishdate":"2019-04-05T00:00:00Z","relpermalink":"/blog/preprintog/","section":"blog","summary":"How many lynx are there? \u0026quot;Spatial density estimates of Eurasian lynx (Lynx lynx) in the 🇫🇷 Jura and Vosges 🏔️\u0026quot; https://t.co/roLlOsFxw7 @oncfs @CROC_Estelle @ZiFridolin w/ awesome #oSCR @rstats 📦 @chrissuthy @andyroyle_pwrc @dwlinden @CNRS_OccitaniE @INEE_CNRS pic.","tags":["SCR","lynx"],"title":"Preprint on how many lynx in France","type":"blog"},{"authors":null,"categories":null,"content":"@FredBarraquand is on 🔥 - check out his new #arXiv preprint \u0026quot;Fitting stochastic predator-prey models using both population density and kill rate data\u0026quot; https://t.co/u23GoSoo81 pic.twitter.com/0YwHWbNepb\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 4 avril 2019 ","date":1554336000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1554336000,"objectID":"e37da58d188f01e24aa480a179e7d3b1","permalink":"https://oliviergimenez.github.io/blog/preprintfred2/","publishdate":"2019-04-04T00:00:00Z","relpermalink":"/blog/preprintfred2/","section":"blog","summary":"@FredBarraquand is on 🔥 - check out his new #arXiv preprint \u0026quot;Fitting stochastic predator-prey models using both population density and kill rate data\u0026quot; https://t.co/u23GoSoo81 pic.twitter.com/0YwHWbNepb\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 4 avril 2019 ","tags":["Democom","multispeces"],"title":"Fitting stochastic predator-prey models using both population density and kill rate data","type":"blog"},{"authors":null,"categories":null,"content":"🥳🧐 New preprint w/ @FredBarraquand \u0026quot;Integrating multiple data sources to fit matrix population models for interacting species\u0026quot; https://t.co/LSSqOcj0Ca w/ codes https://t.co/Msn8XnChu6; check out our #democom project for more https://t.co/eNFOzrPgWR #rstats #IPMs #predatorprey pic.twitter.com/kJdX16rbU6\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 2 avril 2019 ","date":1554163200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1554163200,"objectID":"a40b07f73bdab26d523de99606dc07e2","permalink":"https://oliviergimenez.github.io/blog/preprintfred1/","publishdate":"2019-04-02T00:00:00Z","relpermalink":"/blog/preprintfred1/","section":"blog","summary":"🥳🧐 New preprint w/ @FredBarraquand \u0026quot;Integrating multiple data sources to fit matrix population models for interacting species\u0026quot; https://t.co/LSSqOcj0Ca w/ codes https://t.co/Msn8XnChu6; check out our #democom project for more https://t.co/eNFOzrPgWR #rstats #IPMs #predatorprey pic.","tags":["Democom","ipm","multispeces"],"title":"Integrating multiple data sources to fit matrix population models for interacting species","type":"blog"},{"authors":null,"categories":null,"content":"New preprint by awesome #SarahCubaynes \u0026quot;Modeling the demography of species providing extended parental care: A #capturerecapture approach with a case study on #Polar Bears\u0026quot; https://t.co/x1H0X7GyKn #demography #rstats #multistate #WomenInSTEM #WomenInScience pic.twitter.com/rYEwOUobEn\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 2 avril 2019 ","date":1554163200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1554163200,"objectID":"4887cc190894d3c67d34ab8136da928d","permalink":"https://oliviergimenez.github.io/blog/preprintsarah/","publishdate":"2019-04-02T00:00:00Z","relpermalink":"/blog/preprintsarah/","section":"blog","summary":"New preprint by awesome #SarahCubaynes \u0026quot;Modeling the demography of species providing extended parental care: A #capturerecapture approach with a case study on #Polar Bears\u0026quot; https://t.co/x1H0X7GyKn #demography #rstats #multistate #WomenInSTEM #WomenInScience pic.","tags":["polar bear","demography","hmm"],"title":"Modeling the demography of species providing extended parental care: A capturerecapture approach with a case study on Polar Bears","type":"blog"},{"authors":null,"categories":null,"content":"New paper by #VincenzoGervasi \u0026quot;Failure to coordinate management in transboundary populations hinders the achievement of national management goals: The case of wolverines in Scandinavia\u0026quot; w/ @johnlinnell #HenrikBroseth #LabexCemeb @CNRS_OccitaniE @NINAforskning @umontpellier pic.twitter.com/gzkANotqry\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 2 avril 2019 ","date":1554163200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1554163200,"objectID":"05498fd80749f677327b726d557dec87","permalink":"https://oliviergimenez.github.io/blog/papervincenzo/","publishdate":"2019-04-02T00:00:00Z","relpermalink":"/blog/papervincenzo/","section":"blog","summary":"New paper by #VincenzoGervasi \u0026quot;Failure to coordinate management in transboundary populations hinders the achievement of national management goals: The case of wolverines in Scandinavia\u0026quot; w/ @johnlinnell #HenrikBroseth #LabexCemeb @CNRS_OccitaniE @NINAforskning @umontpellier pic.","tags":["wolverine","demography"],"title":"New paper: Failure to coordinate management in transboundary populations hinders the achievement of national management goals: The case of wolverines in Scandinavia","type":"blog"},{"authors":null,"categories":null,"content":"🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩 @CNRS_OccitaniE @INEE_CNRS @CNRS @CollegeDoctoUM @IsiteMUSE @umontpellier https://t.co/6JlClESHo5\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 13 février 2019 ","date":1550016000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1550016000,"objectID":"d4160f7aeba9fd8a0da334c1e88cd554","permalink":"https://oliviergimenez.github.io/blog/prixlarecherche/","publishdate":"2019-02-13T00:00:00Z","relpermalink":"/blog/prixlarecherche/","section":"blog","summary":"🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩🤩 @CNRS_OccitaniE @INEE_CNRS @CNRS @CollegeDoctoUM @IsiteMUSE @umontpellier https://t.co/6JlClESHo5\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 13 février 2019 ","tags":["La Recherche","wolf"],"title":"Julie obtient le prix du magazine La Recherche","type":"blog"},{"authors":null,"categories":null,"content":"👩 💻🗺️👨 💻 The slides of my introduction to #GIS and #mapping in #rstats using the #sf 📦 and brown 🐻 distribution in the #pyrenees as a case study https://t.co/SKQOCzbxHn - raw material on #github https://t.co/dHoMz6I2Kp #rspatial #spatial #ggplot2 #tidyverse #dataviz pic.twitter.com/22eD1Y55d3\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 8 février 2019 ","date":1549584000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1549584000,"objectID":"121c6a2612fbf0daa64360277c0106e1","permalink":"https://oliviergimenez.github.io/blog/intro_spatial/","publishdate":"2019-02-08T00:00:00Z","relpermalink":"/blog/intro_spatial/","section":"blog","summary":"👩 💻🗺️👨 💻 The slides of my introduction to #GIS and #mapping in #rstats using the #sf 📦 and brown 🐻 distribution in the #pyrenees as a case study https://t.co/SKQOCzbxHn - raw material on #github https://t.","tags":["GIS","R","sf","ggplot","tidyverse"],"title":"Introduction to spatial analyses in R","type":"blog"},{"authors":null,"categories":null,"content":"Meeting of our Interact project https://t.co/Le2mkruYag on the interactions between bottlenose 🐬 and human activities in the French Mediterranean 🌊 @Fondationfrance @GIS_3M #EcoOcéanInstitut #SouffleursdEcume #Gecem @LIENSsCNRSULR @WWFFrance @WWFFrance_Press @CNRS_OccitaniE pic.twitter.com/hAJPnumxAr\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 6 février 2019 ","date":1549411200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1549411200,"objectID":"ec6b39e4ce3f38fa64070cef934e6c0a","permalink":"https://oliviergimenez.github.io/blog/interact_meeting2/","publishdate":"2019-02-06T00:00:00Z","relpermalink":"/blog/interact_meeting2/","section":"blog","summary":"Meeting of our Interact project https://t.co/Le2mkruYag on the interactions between bottlenose 🐬 and human activities in the French Mediterranean 🌊 @Fondationfrance @GIS_3M #EcoOcéanInstitut #SouffleursdEcume #Gecem @LIENSsCNRSULR @WWFFrance @WWFFrance_Press @CNRS_OccitaniE pic.twitter.com/hAJPnumxAr","tags":["meeting","interact"],"title":"Meeting on interactions between dolphins and human activities","type":"blog"},{"authors":null,"categories":null,"content":"Procrastination at its highest level 😜 My first attempt to design hex stickers for our #R2ucare 📦 https://t.co/ZbclwJKB5W #rstats code on #github https://t.co/TbEgjQf7iC Comments more than welcome 😁 pic.twitter.com/vVdVnJ9B79\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 6 février 2019 ","date":1549411200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1549411200,"objectID":"dab433d44c3fe2741e57bb62f7a3cb44","permalink":"https://oliviergimenez.github.io/blog/procrastination/","publishdate":"2019-02-06T00:00:00Z","relpermalink":"/blog/procrastination/","section":"blog","summary":"Procrastination at its highest level 😜 My first attempt to design hex stickers for our #R2ucare 📦 https://t.co/ZbclwJKB5W #rstats code on #github https://t.co/TbEgjQf7iC Comments more than welcome 😁 pic.twitter.com/vVdVnJ9B79\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 6 février 2019 ","tags":["logo","R","R2ucare"],"title":"Procrastination...","type":"blog"},{"authors":null,"categories":null,"content":"How many 🐺🐶 hybrids are there?! @NSantostasi shows how to estimate prevalence w/ #HMM imperfect detection in her @Ecol_Evol #Euring17 paper https://t.co/KwiHB2ZBpX #openaccess 🤝 work w/ @SapienzaRoma @ISPRA_Press @wolfappcenter @umontpellier @CNRS_OccitaniE @INEE_CNRS 🇮🇹🇫🇷 pic.twitter.com/jQJtBwNOZn\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 5 février 2019 ","date":1549324800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1549324800,"objectID":"94bd070e161a2de6acee5fc0709352cf","permalink":"https://oliviergimenez.github.io/blog/paper_nina/","publishdate":"2019-02-05T00:00:00Z","relpermalink":"/blog/paper_nina/","section":"blog","summary":"How many 🐺🐶 hybrids are there?! @NSantostasi shows how to estimate prevalence w/ #HMM imperfect detection in her @Ecol_Evol #Euring17 paper https://t.co/KwiHB2ZBpX #openaccess 🤝 work w/ @SapienzaRoma @ISPRA_Press @wolfappcenter @umontpellier @CNRS_OccitaniE @INEE_CNRS 🇮🇹🇫🇷 pic.","tags":["hybrid prevalence","democom","hmm"],"title":"How to estimate the prevalence of hybrids","type":"blog"},{"authors":null,"categories":null,"content":"Check out our new #bioRxiv preprint on the influence of maternal traits on litter size in #polar #bears led by former Master student #DorindaFolio \u0026quot;How many cubs can a mum raise? Maternal age and size influence litter size in polar bears\u0026quot; https://t.co/kJymSuIkMZ #WomenInSTEM pic.twitter.com/rrCYScxmlj\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 29 janvier 2019 ","date":1548720000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1548720000,"objectID":"7b31e5f2dc952252a31beb03c23dd9a8","permalink":"https://oliviergimenez.github.io/blog/preprint_dorinda/","publishdate":"2019-01-29T00:00:00Z","relpermalink":"/blog/preprint_dorinda/","section":"blog","summary":"Check out our new #bioRxiv preprint on the influence of maternal traits on litter size in #polar #bears led by former Master student #DorindaFolio \u0026quot;How many cubs can a mum raise?","tags":["life-history","polar bear"],"title":"How many cubs can a mum raise? Maternal age and size influence litter size in polar bears","type":"blog"},{"authors":null,"categories":null,"content":"My introduction to the #tidyverse for our lab meeting to manipulate and visualise data in #rstat https://t.co/As9bkXY9GZ. Feel free to steal and modify this material for your own use. Be advised, this is work in progress \u0026amp; a mix of 🇬🇧/🇫🇷😋 Comments welcome! #datascience #davaviz pic.twitter.com/2vrrdbzuWh\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 16 janvier 2019 ","date":1547596800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1547596800,"objectID":"a42109475dbc03270996c768573f491e","permalink":"https://oliviergimenez.github.io/blog/intro_tidyverse/","publishdate":"2019-01-16T00:00:00Z","relpermalink":"/blog/intro_tidyverse/","section":"blog","summary":"My introduction to the #tidyverse for our lab meeting to manipulate and visualise data in #rstat https://t.co/As9bkXY9GZ. Feel free to steal and modify this material for your own use. Be advised, this is work in progress \u0026amp; a mix of 🇬🇧/🇫🇷😋 Comments welcome!","tags":["tidyverse","R","data wrangling"],"title":"Introduction to the Tidyverse","type":"blog"},{"authors":null,"categories":null,"content":"Check out our new @biorxivpreprint #preprint \u0026quot;Inferring #wildlife #poaching in Southeast Asia with multispecies dynamic #occupancy models\u0026quot; led by @LMarescot 🐯 https://t.co/g4NAQnUODw #HMMs @ArnaudLyet @rhohits #NeilCarter @SMARTCnsvTools @AgenceRecherche pic.twitter.com/5EttLyY1S9\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 15 janvier 2019 ","date":1547510400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1547510400,"objectID":"2be01cf2c354354d99c2fc053025fab5","permalink":"https://oliviergimenez.github.io/blog/preprint_poaching/","publishdate":"2019-01-15T00:00:00Z","relpermalink":"/blog/preprint_poaching/","section":"blog","summary":"Check out our new @biorxivpreprint #preprint \u0026quot;Inferring #wildlife #poaching in Southeast Asia with multispecies dynamic #occupancy models\u0026quot; led by @LMarescot 🐯 https://t.co/g4NAQnUODw #HMMs @ArnaudLyet @rhohits #NeilCarter @SMARTCnsvTools @AgenceRecherche pic.twitter.com/5EttLyY1S9\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 15 janvier 2019 ","tags":["democom","carnivores","multispecies","occupancy models","poaching"],"title":"Inferring wildlife poaching in Southeast Asia with multispecies dynamic occupancy models","type":"blog"},{"authors":null,"categories":null,"content":"\u0026quot;Identifying conservation priorities for the critically endangered Balkanlynx\u0026quot; https://t.co/EDx0hUUT7v a nice article by #DimeMelovski on our recent @OryxTheJournal paper https://t.co/sVeTwbDlOc (15 days free access) w/ #rstat #OccupancyModels #GIS #maps https://t.co/Uv9uhsAPuY pic.twitter.com/c3iMLOChU3\n\u0026mdash; Olivier Gimenez 🖖 (@oaggimenez) 14 janvier 2019kquote\u003e ","date":1547424000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1547424000,"objectID":"7b1362c1aa2b09fe41393d6816ba8d3d","permalink":"https://oliviergimenez.github.io/blog/comm_balkanlynx/","publishdate":"2019-01-14T00:00:00Z","relpermalink":"/blog/comm_balkanlynx/","section":"blog","summary":"\u0026quot;Identifying conservation priorities for the critically endangered Balkanlynx\u0026quot; https://t.co/EDx0hUUT7v a nice article by #DimeMelovski on our recent @OryxTheJournal paper https://t.co/sVeTwbDlOc (15 days free access) w/ #rstat #OccupancyModels #GIS #maps https://t.co/Uv9uhsAPuY pic.","tags":["large carnivores","Balkan lynx","occupancy models"],"title":"Identifying conservation priorities for the critically endangered Balkan lynx","type":"blog"},{"authors":null,"categories":null,"content":"In statistical ecology, we often need to calculate the sampling variance of a function of an estimate of which we do know the sampling variance. I keep forgetting how to implement the so-called delta method in R that allows to get an approximation of this quantity. So in this post I go through two examples in population ecology that should help me remembering. I use the deltamethod function from the msm package.\nLoad the package msm and get some help on the delta-method function:\nlibrary(msm) ?deltamethod Further examples can be obtained by typing in:\nexample(deltamethod) For a nice introduction to the delta method, check that out. Two papers are worth reading on the topic: \u0026lsquo;Who Invented the Delta Method?\u0026rsquo; by Jay M. Ver Hoef and \u0026lsquo;Approximating variance of demographic parameters using the delta method: A reference for avian biologists\u0026rsquo; by Larkin A. Powell.\nSimple example A simple example is when, for example, you get $\\phi$ (ignore the traditional hat) an estimate of a survival probability on the logit scale in some capture-recapture analyses, and you would like to get the standard error (SE) of survival on its natural scale.\nFor example, say $\\text{logit}(\\phi) = \\text{lphi} = -0.4473122$ with $\\text{SE} = 0.3362757$.\nTo obtain $\\phi$, you back-transform $\\text{lphi}$ using the reciprocal function of the logit function: $$\\phi = \\displaystyle{\\frac{\\exp(\\text{lphi})}{1+\\exp(\\text{lphi})}} = \\displaystyle{\\frac{1}{1+\\exp(\\text{-lphi})}} = \\displaystyle{\\frac{1}{1+\\exp(\\text{-(-0.4473122)})}} = 0.39.$$\nWhat about the SE of $\\phi$? Well, a direct application of the deltamethod function from the msm package gives the answer:\ndeltamethod(~ 1/(1+exp(-x1)), -0.4473122, 0.3362757^2) ## [1] 0.07999999 Two things to take care of:\nFirst, the variables in the formula must be labelled $x_1, x_2, \\text{text}$. You cannot use $x, y, z, \u0026hellip;$ for example. Just numbered $x$\u0026rsquo;s.\nSecond, the input parameters are the estimate and its squared SE (not the SE), and by default you will get as an output the SE (not the squared SE) of the function defined by the formula.\nComplex example This example deals with an occupancy model. It is a bit more complex than the previous example because we consider a function of several parameters for which we would like to calculate its SE. I assume that occupancy at first occasion was estimated along with its SE, and that one would like to obtain the SE of subsequent occupancy probabilities.\nI calculate time-dependent occupancy probabilities with the following formula $$\\psi_{t+1} = \\psi_t (1 - \\varepsilon) + (1 - \\psi_t) \\gamma$$ where $\\varepsilon$ is extinction, $\\gamma$ is colonisation and $\\psi_t$ is occupancy year $t$.\nWe assume that we obtained the following parameter estimates:\nepsilon = 0.39 gamma = 0.07 psi_init = 0.1 # first-occasion occupancy with corresponding SEs:\nse_epsilon = 0.08 se_psi_init = 0.01 se_gamma = 0.05 We will estimate occupancy and get SEs at 10 occasions, which we store in two matrices (column vectors):\npsi = matrix(0, nrow = 10, ncol = 1) psi_se = matrix(0, nrow = 10, ncol = 1) The first element is occupancy at first occasion:\npsi[1,] \u0026lt;- psi_init psi_se[1,] \u0026lt;- se_psi_init Then we iterate calculations using the formula above:\nfor(i in 2:10){ psi_current \u0026lt;- psi[i-1,] psi_se_current \u0026lt;- psi_se[i-1,] estmean \u0026lt;- c(psi_current,epsilon,gamma) estvar \u0026lt;- diag(c(psi_se_current,se_epsilon,se_gamma)^2) psi[i,] = (psi_current*(1-epsilon)) + ((1-psi_current)*gamma) # recurrence formula psi_se[i,] = deltamethod(~ x1*(1-x2) + (1-x1)*x3, estmean, estvar) # delta-method } Display results:\ndata.frame(psi = psi,sterr_psi = psi_se) ## psi sterr_psi ## 1 0.1000000 0.01000000 ## 2 0.1240000 0.04602347 ## 3 0.1369600 0.05132740 ## 4 0.1439584 0.05244394 ## 5 0.1477375 0.05259904 ## 6 0.1497783 0.05255782 ## 7 0.1508803 0.05250010 ## 8 0.1514753 0.05245886 ## 9 0.1517967 0.05243372 ## 10 0.1519702 0.05241934 Here, we assumed that sampling correlation was 0, in other words that the estimates of $\\psi$, $\\gamma$ and $\\epsilon$ were independent, hence the use of a diagonal matrix for estvar. It is possible to use a non-diagonal covariance matrix to account for non-null correlation.\n","date":1546560000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1546560000,"objectID":"7e0e44064021c26712608a491a4af7e7","permalink":"https://oliviergimenez.github.io/blog/delta-method/","publishdate":"2019-01-04T00:00:00Z","relpermalink":"/blog/delta-method/","section":"blog","summary":"In statistical ecology, we often need to calculate the sampling variance of a function of an estimate of which we do know the sampling variance. I keep forgetting how to implement the so-called delta method in R that allows to get an approximation of this quantity. So in this post I go through two examples in population ecology that should help me remembering. I use the deltamethod function from the msm package.\n","tags":["statistical ecology","estimation","delta-method","rstats"],"title":"Calculate the standard error of any function using the delta method","type":"blog"},{"authors":null,"categories":null,"content":"I read this awesome post (in French) by Baptiste Coulmont, professor in sociology, who explored the French academic network in sociology. Coulmont used the composition of PhD commitees to determine academic links between colleagues. The approach very appealing because it uses public data available from the website these.fr. Here, I used Coulmont\u0026rsquo;s R code to produce the French academic network in ecology. This was a nice opportunity to illustrate how to work in the tidyverse and to do some web scraping using the rvest package.\nGet the data Load the packages we need:\nlibrary(RCurl) library(tidyverse) library(lubridate) library(scales) library(hrbrthemes) library(data.table) # devtools::install_github(\u0026quot;privefl/bigreadr\u0026quot;) library(bigreadr) We now prepare the URL requests. The total number of PhDs is around 88000 on the period 2015-2018. Because the website uses slices of 1000 on each page, we proceed in sequence:\ni \u0026lt;- 1:88 i \u0026lt;- i*1000 URL \u0026lt;-paste0(\u0026quot;http://www.theses.fr/?q=\u0026amp;fq=dateSoutenance:[2015-01-01T23:59:59Z%2BTO%2B2018-12-31T23:59:59Z]\u0026amp;checkedfacets=\u0026amp;start=\u0026quot;,i,\u0026quot;\u0026amp;sort=none\u0026amp;status=\u0026amp;access=\u0026amp;prevision=\u0026amp;filtrepersonne=\u0026amp;zone1=titreRAs\u0026amp;val1=\u0026amp;op1=AND\u0026amp;zone2=auteurs\u0026amp;val2=\u0026amp;op2=AND\u0026amp;zone3=etabSoutenances\u0026amp;val3=\u0026amp;zone4=dateSoutenance\u0026amp;val4a=\u0026amp;val4b=\u0026amp;type=\u0026amp;lng=\u0026amp;checkedfacets=\u0026amp;format=csv\u0026quot;) Alternatively, the search can be done by hand directly from the theses.fr website. [François-Xavier Coudert] ( https://www.coudert.name/) also provides the search results for the 2015-2018 period.\nWe proceed with the requests, and store everything in a csv file:\nj \u0026lt;- 1 SERP \u0026lt;- 1 for(j in 1:length(URL)){ # loop over the slices SERP[j] \u0026lt;- getURL(URL[j]) write.csv(SERP,\u0026quot;SERP_2.csv\u0026quot;,append=F) } rm(SERP,i,j,URL) We keep only the PhDs in the field (Discipline) of ecology. This is basically the only change I have made to Coulmont\u0026rsquo;s neat code.\ntheses \u0026lt;- read.csv(\u0026quot;SERP_2.csv\u0026quot;,sep=\u0026quot;;\u0026quot;,quote=\u0026quot;\u0026quot;,skip=1,stringsAsFactors = F) #theses %\u0026gt;% # pull(X..Discipline..) %\u0026gt;% # unique() ecology \u0026lt;- theses %\u0026gt;% filter(grepl(\u0026quot;ecologie\u0026quot;,X..Discipline..,ignore.case=T)) %\u0026gt;% # keep PhDs with Displine == ecologie filter(X..Date.de.soutenance..!=\u0026quot;\u0026quot;) %\u0026gt;% # remove PhDs with missing dates of defense filter(X..Statut..==\u0026quot;soutenue\u0026quot;) # keep only PhDs that have been defended We now have the id of all PhDs in ecology defended during the period 2015-2018. We will use the id to get the composition of all PhD commitees. Getting this composition requires scraping the web page of each PhD, and to get the ID of each PhD. For doing so, we use the rvest package (see the excellent posts by Maëlle Salmon for examples).\nlibrary(rvest) identifiants \u0026lt;- ecology$X..Identifiant.de.la.these.. # get PhD ids reseau_total \u0026lt;- data_frame(noms_jury=\u0026quot;\u0026quot;, liens_jury=\u0026quot;\u0026quot;, these=\u0026quot;\u0026quot;, directeurs=\u0026quot;\u0026quot;, liens_directeurs=\u0026quot;\u0026quot;) for (i in 1:length(identifiants)) { # get info on current PhD data_theses_eco \u0026lt;- read_html( paste0(\u0026quot;http://www.theses.fr/\u0026quot;,identifiants[i]) ) # get name PhD supervisor for directeurs \u0026lt;- bind_cols( directeurs = data_theses_eco %\u0026gt;% html_nodes(\u0026quot;div .donnees-ombre p\u0026quot;) %\u0026gt;% .[[1]] %\u0026gt;% html_nodes(\u0026quot;a\u0026quot;) %\u0026gt;% html_text() , liens_directeurs = data_theses_eco %\u0026gt;% html_nodes(\u0026quot;div .donnees-ombre p\u0026quot;) %\u0026gt;% .[[1]] %\u0026gt;% html_nodes(\u0026quot;a\u0026quot;) %\u0026gt;% html_attr(name=\u0026quot;href\u0026quot;) ) %\u0026gt;% mutate( these = identifiants[i] ) # get names of people in commitees jury \u0026lt;- bind_cols( noms_jury = data_theses_eco %\u0026gt;% html_nodes(\u0026quot;div .donnees p a\u0026quot;) %\u0026gt;% html_text() , liens_jury = data_theses_eco %\u0026gt;% html_nodes(\u0026quot;div .donnees p a\u0026quot;) %\u0026gt;% html_attr(name=\u0026quot;href\u0026quot;) ) %\u0026gt;% mutate( these = identifiants[i] ) # put all together reseau \u0026lt;- jury %\u0026gt;% left_join(directeurs,by=\u0026quot;these\u0026quot;) reseau_total \u0026lt;- bind_rows(reseau_total,reseau) } Build the network Load the packages we need, and the data we got at the previous step:\nlibrary(igraph) library(ggraph) library(ggrepel) load('reseau_total.RData') Coulmont defined a weighted link between two colleagues i and j as follows: 3 if i and j are both supervisors, 2 if i is a supervisor and j a PhD commitee member and 1 if both i and j are PhD commitee members. A colleague may accumulate several weights.\ndirections_theses \u0026lt;- reseau_total %\u0026gt;% select(these,directeurs) %\u0026gt;% unique() %\u0026gt;% group_by(these) %\u0026gt;% mutate(N=n()) %\u0026gt;% filter(N==2) %\u0026gt;% # keep co-supervision w/ 2 supervisors mutate(rang=rank(directeurs)) %\u0026gt;% spread(key=rang,value=directeurs) %\u0026gt;% ungroup() %\u0026gt;% select(nom1=`1`,nom2=`2`) %\u0026gt;% mutate(poids=3) directions_jury \u0026lt;- reseau_total %\u0026gt;% select(nom1=noms_jury,nom2=directeurs) %\u0026gt;% filter( nom1 != \u0026quot;\u0026quot;) %\u0026gt;% mutate(poids=2) %\u0026gt;% group_by(nom1,nom2) %\u0026gt;% summarize(poids=sum(poids)) jury_jury \u0026lt;- reseau_total %\u0026gt;% select(noms_jury,these) %\u0026gt;% unique() %\u0026gt;% filter(noms_jury!=\u0026quot;\u0026quot;) g_j \u0026lt;- graph_from_data_frame(jury_jury,directed=F) V(g_j)$type \u0026lt;- V(g_j)$name %in% jury_jury$noms_jury g_j_1 \u0026lt;- bipartite_projection(g_j,which=\u0026quot;true\u0026quot;) jurys \u0026lt;- as_long_data_frame(g_j_1) %\u0026gt;% select(nom1=`ver[el[, 1], ]`, nom2=`ver2[el[, 2], ]`, poids=weight) reseau_petit \u0026lt;- bind_rows(directions_theses,directions_jury,jurys) %\u0026gt;% group_by(nom1,nom2) %\u0026gt;% summarize(poids=sum(poids)) # data.frame from which the network will be created Each node in the network has a size proportional to its betweenness score. We also determine communities using the walktrap algorithm that will be colored differently. The width of an edge is proportional to the strength of the link between the two corresponding nodes.\ng \u0026lt;- graph_from_data_frame(reseau_petit, directed=F) # create network from data.frame g \u0026lt;- simplify(g,edge.attr.comb = sum) V(g)$degres \u0026lt;- degree(g) V(g)$label \u0026lt;- gsub(\u0026quot;^\\\\S+\\\\s+(.+)$\u0026quot;,\u0026quot;\\\\1\u0026quot;,V(g)$name) V(g)$communaute \u0026lt;- as.character(cluster_walktrap(g, steps=15)$membership) # determine communities V(g)$closeness \u0026lt;- (5*closeness(g))^10 V(g)$btwns \u0026lt;- betweenness(g) # network metric betweeness V(g)$eigen_centr \u0026lt;- eigen_centrality(g)$vector g \u0026lt;- delete_edges(g, which(E(g)$poids\u0026lt;5) ) # delete edges with weight \u0026lt;= 4 V(g)$cluster_number \u0026lt;- clusters(g)$membership # to which community you belong g \u0026lt;- induced_subgraph(g, V(g)$cluster_number== which( max(clusters(g)$csize) == clusters(g)$csize) ) E(g)$weight \u0026lt;- 1/E(g)$poids # width of edge proportional to weight V(g)$label \u0026lt;- ifelse(V(g)$degres\u0026lt;20,\u0026quot;\u0026quot;,V(g)$label) # do not display all names Plot the network We now plot the network. For clarity, we only indicate the names of colleagues who were part of several phD commitees.\nggraph(g,layout=\u0026quot;igraph\u0026quot;,algorithm=\u0026quot;fr\u0026quot;) + geom_edge_link(aes(width=.1*poids), alpha=.1, end_cap = circle(5, 'mm'), start_cap = circle(5, 'mm')) + geom_node_point(aes(size=eigen_centr), color=\u0026quot;white\u0026quot;,alpha=1) + geom_node_point(aes(color=communaute,size=eigen_centr), alpha=.5) + scale_size_area(max_size = 20) + geom_node_text(aes(label=label),size=3,repel=T,box.padding = 0.15) + labs(title=\u0026quot;Réseaux des écologues\u0026quot;, subtitle=\u0026quot;Soutenances de thèses entre 2015 et 2018\u0026quot;, caption=\u0026quot;Sources : theses.fr \\n Code par B. Coulmont, modifié par O. Gimenez\u0026quot;) + theme_graph(foreground = 'white', fg_text_colour = 'white', base_family = \u0026quot;Helvetica\u0026quot;) + theme(legend.position=\u0026quot;none\u0026quot;, text=element_text(size=16,family=\u0026quot;Helvetica\u0026quot;), plot.margin = unit(c(0.2, 0.2, 0.2, 0.2), units=\u0026quot;line\u0026quot;)) # save ggsave(filename = \u0026quot;ecology_network.pdf\u0026quot;,width=30,height = 20) I played around the defaults Coulmont used to build and plot the network. It helps in getting a better understanding of the network and the links between colleagues working in ecology. Overall, I indeed feel very much connected to my colleagues in Montpellier, Lyon and Grenoble. I should probably go out of my comfort zone and interact even more with my colleagues from La Rochelle, Marseille and Aix-en-Provence 😃\nAs always, data and code are available from GitHub.\n","date":1546387200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1546387200,"objectID":"8cc6325e87401f224bee7aedb8164272","permalink":"https://oliviergimenez.github.io/blog/network_ecology/","publishdate":"2019-01-02T00:00:00Z","relpermalink":"/blog/network_ecology/","section":"blog","summary":"I read this awesome post (in French) by Baptiste Coulmont, professor in sociology, who explored the French academic network in sociology. Coulmont used the composition of PhD commitees to determine academic links between colleagues. The approach very appealing because it uses public data available from the website these.fr. Here, I used Coulmont\u0026rsquo;s R code to produce the French academic network in ecology. This was a nice opportunity to illustrate how to work in the tidyverse and to do some web scraping using the rvest package.\n","tags":["network","ecology","tidyverse","rstats"],"title":"Scientific research is all about networking","type":"blog"},{"authors":null,"categories":null,"content":"Our team has had a fantastic 2018 year! Inspired by @Maestrelab, below is a thread of some highlights 🍾😜🤩 @CNRS_OccitaniE @INEE_CNRS @umontpellier @IsiteMUSE @LMarescot @NSantostasi @MaudQueroue @ValentinLauret @oksanagrente @GueryLorelei @roques_seb https://t.co/FaP9o0bOD4.\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 23 décembre 2018 \u0026lt;\nblockquote class=\u0026ldquo;twitter-tweet\u0026rdquo; data-lang=\u0026ldquo;fr\u0026rdquo;\u0026gt;1. We published papers in #statistalecology \u0026amp; large mammals #conservation w/ interest in human \u0026amp; social sciences https://t.co/Xlh6reUUQc @Oikos_Journal @PopulEcology @AnimalEcology @JOSS_TheOJ @OryxTheJournal @EcographyJourna @CommsBio @FunEcology @MethodsEcolEvol @ConBiology. pic.twitter.com/FvakGRzacx\n— Olivier Gimenez 💤 (@oaggimenez) 23 décembre 2018\n2. We had amazing interns! D. Folio analysed maternal traits \u0026amp; polar bears reproduction w/ @SCubi25 \u0026amp; J. Aars @NorskPolar; A. Hendrick studied ethnology \u0026amp; whale-whatching w/ N. Lescureux \u0026amp; H. Labach @GIS_3M; @ValentinLauret developed adaptive monitoring for a #MPAs network. pic.twitter.com/7RCc9hAXgm\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 23 décembre 2018 3. Gilles Maurer and @LouvrierJulie successfully defended their PhD on the the interactions between captive \u0026amp; wild pops of 🐘 w/ @zoobeauval and the statistical #modelling of large #carnivores 🐺🐱 in Europe w/ @ONCFS_officiel \u0026amp; @umontpellier @CollegeDoctoUM. pic.twitter.com/rAtJifrTb6\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 23 décembre 2018 4. We engaged w/ stakeholders to try and have an impact on wildlife management. Several projects are on tracks in that direction https://t.co/R4UcwpDxVN. pic.twitter.com/xovnHc6HCq\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 23 décembre 2018 5. We had awesome visitors (remotely or not) who contributed to the #boulet spirit. Looking in your direction @AmandaVincent1, @u_arbieu, C. Oosthuizen, @pardo_deborah, @saveourseas, @StatnMap, @FredBarraquand, @MoonbeamLevels, @BenhaiemSarah, @CROC_Estelle and others I forget. pic.twitter.com/nF12RjpZrs\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 23 décembre 2018 6. New adventures for 2019 include embarking in 3 new PhDs: @ValentinLauret on 🐬 #adaptive #monitoring in #MPAs w/ @GIS_3M, @MaudQueroue on multispecies 🐦 #demography w/ #CEBC @UMR_MECADEV and @oksanagrente on 🐺 #adaptive #management w/ @ONCFS_officiel @chamaillejammes. pic.twitter.com/nCGfaruLIf\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 23 décembre 2018 7. We also spent time planning the future, and I can tell you, 2019 is gonna be even more awesome than 2018 w/ our new team on the interactions between humans and wildlife #interdisciplinarity #management #shs @abesnardEPHE @INEE_CNRS @inshs_cnrs @CNRS_OccitaniE. pic.twitter.com/3TEAUzC5Ho\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 23 décembre 2018 We https://t.co/FaP9o0bOD4 wish you Merry Christmas and Happy New Year, all the best and see you in 2019 🎄🤶🎅\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 23 décembre 2018 ","date":1546041600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1546041600,"objectID":"39ffc16e5d0f63244c2906d93acb5d62","permalink":"https://oliviergimenez.github.io/blog/2018retrospective/","publishdate":"2018-12-29T00:00:00Z","relpermalink":"/blog/2018retrospective/","section":"blog","summary":"Our team has had a fantastic 2018 year! Inspired by @Maestrelab, below is a thread of some highlights 🍾😜🤩 @CNRS_OccitaniE @INEE_CNRS @umontpellier @IsiteMUSE @LMarescot @NSantostasi @MaudQueroue @ValentinLauret @oksanagrente @GueryLorelei @roques_seb https://t.co/FaP9o0bOD4.\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 23 décembre 2018 \u0026lt;\n","tags":["retrospective","boulet team"],"title":"Looking back at 2018","type":"blog"},{"authors":null,"categories":null,"content":"Pretty proud of my modest contribution: data and #rstats code https://t.co/EUG4IgPdKR 🤓\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 3 décembre 2018 Out now: The Balkan #lynx is one of the rarest, most threatened \u0026amp; least-studied large carnivores. Melovski et al. used questionnaire surveys \u0026amp; occupancy modelling to identify conservation priorities for this Critically Endangered subspecies.https://t.co/asollm8Ior pic.twitter.com/af6jTFd9EI\n\u0026mdash; Oryx (@OryxTheJournal) 5 décembre 2018 ","date":1543795200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1543795200,"objectID":"09dd38e35b5f5080d9bbdf4dd66b155d","permalink":"https://oliviergimenez.github.io/blog/articlebalkanlynx/","publishdate":"2018-12-03T00:00:00Z","relpermalink":"/blog/articlebalkanlynx/","section":"blog","summary":"Pretty proud of my modest contribution: data and #rstats code https://t.co/EUG4IgPdKR 🤓\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 3 décembre 2018 Out now: The Balkan #lynx is one of the rarest, most threatened \u0026amp; least-studied large carnivores.","tags":["balkan lynx","occupancy models","conservation"],"title":"New paper on the Balkan lynx","type":"blog"},{"authors":null,"categories":null,"content":"Check out our @CommsBio paper w/ @IZWberlin \u0026quot;Slow recovery from a disease #epidemic in the #spotted #hyena, a keystone social #carnivore\u0026quot; led by #SarahBenhaiem \u0026amp; @LMarescot @CNRS_OccitaniE @INEE_CNRS @IsiteMUSE @umontpellier #matrixmodel #capturerecapture pic.twitter.com/kangaJYtha\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 212018 ","date":1542758400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1542758400,"objectID":"b2f5501d98dfad3551832c922ee4d6ce","permalink":"https://oliviergimenez.github.io/blog/articlesarahetlulu/","publishdate":"2018-11-21T00:00:00Z","relpermalink":"/blog/articlesarahetlulu/","section":"blog","summary":"Check out our @CommsBio paper w/ @IZWberlin \u0026quot;Slow recovery from a disease #epidemic in the #spotted #hyena, a keystone social #carnivore\u0026quot; led by #SarahBenhaiem \u0026amp; @LMarescot @CNRS_OccitaniE @INEE_CNRS @IsiteMUSE @umontpellier #matrixmodel #capturerecapture pic.","tags":["hyenas","matrix models","disease ecology"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"Thanks for your visit @u_arbieu, the great talk and the stimulating discussions! Awesome collaborations to come 🤝🍻 🇩🇪🇫🇷🇮🇹🐺 https://t.co/04goJ5iMAI @NSantostasi @LouvrierJulie @roques_seb @oksanagrente @MaudQueroue https://t.co/exZsi2qxOA\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 16ovembre 2018 ","date":1542326400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1542326400,"objectID":"d8f6f87482f258a5f6a6af600659f58d","permalink":"https://oliviergimenez.github.io/blog/ugoarbieu_visit/","publishdate":"2018-11-16T00:00:00Z","relpermalink":"/blog/ugoarbieu_visit/","section":"blog","summary":"Thanks for your visit @u_arbieu, the great talk and the stimulating discussions! Awesome collaborations to come 🤝🍻 🇩🇪🇫🇷🇮🇹🐺 https://t.co/04goJ5iMAI @NSantostasi @LouvrierJulie @roques_seb @oksanagrente @MaudQueroue https://t.co/exZsi2qxOA\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 16ovembre 2018 ","tags":["large carnivores","wolf","human-wildlife conflicts"],"title":"Ugo Arbieu's visit","type":"blog"},{"authors":null,"categories":null,"content":"🐬🇬🇷🇮🇹🇫🇷 New paper by @NSantostasi @INEE_CNRS @CNRS_OccitaniE @IsiteMUSE @umontpellier 🤩👏 https://t.co/6vQ6d9HevV\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 9 novembre 2018 ","date":1541721600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1541721600,"objectID":"cd47ba865110597cd03cd99fcde68978","permalink":"https://oliviergimenez.github.io/blog/articlenina/","publishdate":"2018-11-09T00:00:00Z","relpermalink":"/blog/articlenina/","section":"blog","summary":"🐬🇬🇷🇮🇹🇫🇷 New paper by @NSantostasi @INEE_CNRS @CNRS_OccitaniE @IsiteMUSE @umontpellier 🤩👏 https://t.co/6vQ6d9HevV\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 9 novembre 2018 ","tags":["dolphin","R","rstats","PVA","Greece"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"#INLA workshop on spatio-temporal models in the beautiful city of #Avignon 🤩 #RESSTE #GdREcoStat @oksanagrente @CREEM_cake pic.twitter.com/nuLhIkMiwO\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 718 ","date":1541462400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1541462400,"objectID":"cdb3ab92fd571bd8b2dec3589c64083d","permalink":"https://oliviergimenez.github.io/blog/inla_workshop/","publishdate":"2018-11-06T00:00:00Z","relpermalink":"/blog/inla_workshop/","section":"blog","summary":"#INLA workshop on spatio-temporal models in the beautiful city of #Avignon 🤩 #RESSTE #GdREcoStat @oksanagrente @CREEM_cake pic.twitter.com/nuLhIkMiwO\n\u0026mdash; Olivier Gimenez 💤 (@oaggimenez) 718 ","tags":["INLA","R","rstats","spatio-temporal models","point process"],"title":"Workshop on spatio-temporal models with INLA","type":"blog"},{"authors":null,"categories":null,"content":"The famous #NetLogo butterfly example coded in #rstats w/ #NetLogoR pic.twitter.com/VbUUa5vIep\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 5 novembre 2018 Wolf model from Marucco \u0026amp; @eliotmcintire coded in #rstats w/ #NetLogoR \u0026amp; #SpaDES https://t.co/ERx71CIgZv pic.twitter.com/f0YTHuKGSS\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 5 novembre 2018 ","date":1541376000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1541376000,"objectID":"c63044ebe62464f697477c23a120aa89","permalink":"https://oliviergimenez.github.io/blog/crashcourse_netlogor/","publishdate":"2018-11-05T00:00:00Z","relpermalink":"/blog/crashcourse_netlogor/","section":"blog","summary":"The famous #NetLogo butterfly example coded in #rstats w/ #NetLogoR pic.twitter.com/VbUUa5vIep\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 5 novembre 2018 Wolf model from Marucco \u0026amp; @eliotmcintire coded in #rstats w/ #NetLogoR \u0026amp; #SpaDES https://t.","tags":["IBM","R","rstats","modelling"],"title":"Crash course on individual-based models using NetLogoR","type":"blog"},{"authors":null,"categories":null,"content":"🍃📊 La @sfecologie créé avec le #GDR #ecostat un groupe thématique #écologie #statistique https://t.co/Lv33Ur1POc - site du GDR https://t.co/kTXpjJXlya \u0026amp; liste diffusion https://t.co/y3s0xgyzEJ - plus sur l\u0026#39;écologie statistique https://t.co/aOvTJFGDLu \u0026amp; https://t.co/7TRCKNUMKi pic.twitter.com/C4ifYZ48QC\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 4 novembre 2018 ","date":1541203200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1541203200,"objectID":"4b6e42cd605903d609a47e00ced9022e","permalink":"https://oliviergimenez.github.io/blog/groupeecostatsfee/","publishdate":"2018-11-03T00:00:00Z","relpermalink":"/blog/groupeecostatsfee/","section":"blog","summary":"🍃📊 La @sfecologie créé avec le #GDR #ecostat un groupe thématique #écologie #statistique https://t.co/Lv33Ur1POc - site du GDR https://t.co/kTXpjJXlya \u0026amp; liste diffusion https://t.co/y3s0xgyzEJ - plus sur l\u0026#39;écologie statistique https://t.co/aOvTJFGDLu \u0026amp; https://t.","tags":["book","write"],"title":"Groupe thématique Ecologie Statistique de la SFEE","type":"blog"},{"authors":null,"categories":null,"content":"🐘🌏 Sauver l\u0026#39;éléphant d\u0026#39;Asie avec des congés maternités? Itw de Gilles Maurer dans le magazine #LUM de l\u0026#39;@umontpellier https://t.co/sKEJc8FsJH @IsiteMUSE @CNRSenLR @INEE_CNRS @zoobeauval @UWyonews - article scientifique https://t.co/uqqPfPDvpp et 🎦 https://t.co/1m9YAoe3Q3 pic.twitter.com/xxMxFjtrKS\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 4 novembre 2018 ","date":1541030400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1541030400,"objectID":"164a183433e0151de3144fb916a5fac3","permalink":"https://oliviergimenez.github.io/blog/howtowritealot/","publishdate":"2018-11-01T00:00:00Z","relpermalink":"/blog/howtowritealot/","section":"blog","summary":"🐘🌏 Sauver l\u0026#39;éléphant d\u0026#39;Asie avec des congés maternités? Itw de Gilles Maurer dans le magazine #LUM de l\u0026#39;@umontpellier https://t.co/sKEJc8FsJH @IsiteMUSE @CNRSenLR @INEE_CNRS @zoobeauval @UWyonews - article scientifique https://t.co/uqqPfPDvpp et 🎦 https://t.","tags":["vulgarisation","Asian elephant"],"title":"Sauver l'éléphant d'Asie avec des congés maternités?","type":"blog"},{"authors":null,"categories":null,"content":"Nice read \u0026quot;How to write a lot\u0026quot; by Paul Silvia: \u0026quot;Make a writing schedule and show up for it. Want less and do more\u0026quot;, \u0026quot;Productive writing involves harnessing the power of habit, and habits come from repetition\u0026quot; Let\u0026#39;s get to it then! https://t.co/TTAOrFl6Z5 https://t.co/KXgGCq9R2v pic.twitter.com/Eno5qv3g0w\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 2 novembre 201 ","date":1540857600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1540857600,"objectID":"8a285862acdea500036f71167b3da327","permalink":"https://oliviergimenez.github.io/blog/articlelumgilles/","publishdate":"2018-10-30T00:00:00Z","relpermalink":"/blog/articlelumgilles/","section":"blog","summary":"Nice read \u0026quot;How to write a lot\u0026quot; by Paul Silvia: \u0026quot;Make a writing schedule and show up for it. Want less and do more\u0026quot;, \u0026quot;Productive writing involves harnessing the power of habit, and habits come from repetition\u0026quot; Let\u0026#39;s get to it then!","tags":["book","write"],"title":"How to write a lot?","type":"blog"},{"authors":null,"categories":null,"content":"Check out our new preprint by awesome Anne-Sophie Bonnet-Lebrun \u0026quot;Identifying priority conservation areas for recovering large #carnivores using #citizenscience data\u0026quot; w/ 🐟 point-process models to map 🇬🇷 🐻 distribution https://t.co/CEUbVKuXRQ starring @Arcturosgr @MoonbeamLevels pic.twitter.com/fb9UAP3Aep\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 29 oct8 ","date":1540771200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1540771200,"objectID":"a96ec3dd3f50f67a57a165ab894d5345","permalink":"https://oliviergimenez.github.io/blog/newpreprint/","publishdate":"2018-10-29T00:00:00Z","relpermalink":"/blog/newpreprint/","section":"blog","summary":"Check out our new preprint by awesome Anne-Sophie Bonnet-Lebrun \u0026quot;Identifying priority conservation areas for recovering large #carnivores using #citizenscience data\u0026quot; w/ 🐟 point-process models to map 🇬🇷 🐻 distribution https://t.co/CEUbVKuXRQ starring @Arcturosgr @MoonbeamLevels pic.","tags":["sdm","large carnivores","brown bear","point process"],"title":"New preprint!","type":"blog"},{"authors":null,"categories":null,"content":"New interdisciplinary project combining ecological, human and social sciences for large #carnivores management 🐺🐻🐱 https://t.co/04goJ5iMAI @INEE_CNRS @inshs_cnrs @CNRSenLR @ONCFS_officiel @univpaulvalery @CNRS @PACTE_grenoble pic.twitter.com/7IVzJR7l8y\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 17 octobre 2018 Nouveau projet interdisciplinaire combinant sciences écologiques, humaines et sociales pour la gestion des grands #carnivores 🐺🐻🐱 https://t.co/04goJ5iMAI @INEE_CNRS @inshs_cnrs @CNRSenLR @ONCFS_officiel @univpaulvalery @CNRS @PACTE_grenoble pic.twitter.com/BP8rVPkvtD\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 16 ocobre 2018 ","date":1538956800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1538956800,"objectID":"91211eecddc39b58dcaa7a64c69a4089","permalink":"https://oliviergimenez.github.io/blog/ecosocar_website/","publishdate":"2018-10-08T00:00:00Z","relpermalink":"/blog/ecosocar_website/","section":"blog","summary":"New interdisciplinary project combining ecological, human and social sciences for large #carnivores management 🐺🐻🐱 https://t.co/04goJ5iMAI @INEE_CNRS @inshs_cnrs @CNRSenLR @ONCFS_officiel @univpaulvalery @CNRS @PACTE_grenoble pic.twitter.com/7IVzJR7l8y\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 17 octobre 2018 Nouveau projet interdisciplinaire combinant sciences écologiques, humaines et sociales pour la gestion des grands #carnivores 🐺🐻🐱 https://t.","tags":["interdisciplinarity","large carnivores","ecosocar"],"title":"New project!","type":"blog"},{"authors":null,"categories":null,"content":"Meeting about the projet of our new team on the ecology of interactions between humans and wildlife #interdisciplinarity #modelling #management #geography #ethnology #shs #teamboulet @umontpellier @IsiteMUSE @CNRSenLR @INEE_CNRS @inshs_cnrs pic.twitter.com/Vj3OqIkFr6\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 8 tobre 2018 ","date":1538956800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1538956800,"objectID":"b90e244256159d659716777d0c8f9508","permalink":"https://oliviergimenez.github.io/blog/newteam/","publishdate":"2018-10-08T00:00:00Z","relpermalink":"/blog/newteam/","section":"blog","summary":"Meeting about the projet of our new team on the ecology of interactions between humans and wildlife #interdisciplinarity #modelling #management #geography #ethnology #shs #teamboulet @umontpellier @IsiteMUSE @CNRSenLR @INEE_CNRS @inshs_cnrs pic.twitter.com/Vj3OqIkFr6","tags":["interdisciplinarity","team"],"title":"Team meeting","type":"blog"},{"authors":null,"categories":null,"content":"We had fun 😉 pic.twitter.com/C93lSMzts8\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 7octobre 2018 ","date":1538524800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1538524800,"objectID":"91a2ffd4067944ed9c7fb602085bf1e1","permalink":"https://oliviergimenez.github.io/blog/4thmeetinggdrecostat/","publishdate":"2018-10-03T00:00:00Z","relpermalink":"/blog/4thmeetinggdrecostat/","section":"blog","summary":"We had fun 😉 pic.twitter.com/C93lSMzts8\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 7octobre 2018 ","tags":["statistical ecology","GDR"],"title":"4th meeting of the GDR EcoStat","type":"blog"},{"authors":null,"categories":null,"content":"More details here https://t.co/FaP9o0bOD4\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 1 octobre 2018 ","date":1538352000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1538352000,"objectID":"16d33562bff7277d69b5ad6feb653e60","permalink":"https://oliviergimenez.github.io/blog/new_phds/","publishdate":"2018-10-01T00:00:00Z","relpermalink":"/blog/new_phds/","section":"blog","summary":"More details here https://t.co/FaP9o0bOD4\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 1 octobre 2018 ","tags":["phd","boulet"],"title":"Lucky to have 3 new PhD students","type":"blog"},{"authors":null,"categories":null,"content":"🤩🍾🥳 Well done @LouvrierJulie - Impressiiiiiiiiiiiiive work! https://t.co/wd99KZcYTy\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 30 septembre 2018 ","date":1538265600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1538265600,"objectID":"b777688efbb0c5644da2f06511a79dc7","permalink":"https://oliviergimenez.github.io/blog/phdjulie_delivered/","publishdate":"2018-09-30T00:00:00Z","relpermalink":"/blog/phdjulie_delivered/","section":"blog","summary":"🤩🍾🥳 Well done @LouvrierJulie - Impressiiiiiiiiiiiiive work! https://t.co/wd99KZcYTy\n\u0026mdash; Olivier Gimenez 🚸 (@oaggimenez) 30 septembre 2018 ","tags":["large carnivores","sdm","phd","boulet","wolf","lynx","occupancy"],"title":"Julie Louvrier submitted her PhD manuscript!","type":"blog"},{"authors":null,"categories":null,"content":"🐬👩 💻 New paper by awesome @NSantostasi showing that \u0026quot;Common dolphins in the Gulf of Corinth are Critically Endangered\u0026quot; @IUCNRedList 🇬🇷🇮🇹🇫🇷 #pva #conservation @marinemammalogy https://t.co/lcQpyoetm8 #womeninSTEM pic.twitter.com/21raWOutfM\n\u0026mdash; Olivier Gimenez 😴 (@oaggimenez) 30 septembre 2018 ","date":1538265600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1538265600,"objectID":"c7dfbd5504a27ae4697c62d7b37a9a1e","permalink":"https://oliviergimenez.github.io/blog/nina_pva_commondolphin/","publishdate":"2018-09-30T00:00:00Z","relpermalink":"/blog/nina_pva_commondolphin/","section":"blog","summary":"🐬👩 💻 New paper by awesome @NSantostasi showing that \u0026quot;Common dolphins in the Gulf of Corinth are Critically Endangered\u0026quot; @IUCNRedList 🇬🇷🇮🇹🇫🇷 #pva #conservation @marinemammalogy https://t.co/lcQpyoetm8 #womeninSTEM pic.twitter.com/21raWOutfM\n\u0026mdash; Olivier Gimenez 😴 (@oaggimenez) 30 septembre 2018 ","tags":["common dolphins","pva","conservation","cetaceans","dolphin biology and conservation","Greece"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"🤩🐺 New paper by unstoppable @LouvrierJulie \u0026quot;Accounting for #misidentification and #heterogeneity in #occupancy studies using #HMMs\u0026quot; https://t.co/UqJfbkFjbW #teamboulet #openaccess @EcoHMM #carnivorepapersbywomen @ONCFS_officiel @umontpellier @IsiteMUSE @CNRSenLR @INEE_CNRS pic.twitter.com/YGgPNjzRUE\n\u0026mdash; Olivier Gimenez 🥐 (@oaggimenez) 19 septembre18 ","date":1537315200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1537315200,"objectID":"dc432ba16181b9989c1f73e044ff9f6b","permalink":"https://oliviergimenez.github.io/blog/julie_ecolmodel_paper/","publishdate":"2018-09-19T00:00:00Z","relpermalink":"/blog/julie_ecolmodel_paper/","section":"blog","summary":"🤩🐺 New paper by unstoppable @LouvrierJulie \u0026quot;Accounting for #misidentification and #heterogeneity in #occupancy studies using #HMMs\u0026quot; https://t.co/UqJfbkFjbW #teamboulet #openaccess @EcoHMM #carnivorepapersbywomen @ONCFS_officiel @umontpellier @IsiteMUSE @CNRSenLR @INEE_CNRS pic.twitter.com/YGgPNjzRUE\n\u0026mdash; Olivier Gimenez 🥐 (@oaggimenez) 19 septembre18 ","tags":["HMM","occupancy","misidentification","false positives","carnivores"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"New paper by #SarahBenhaiem \u0026amp; @LMarescot 🤩👏 \u0026quot;Robustness of Eco-Epidemiological Capture-Recapture Parameter Estimates to Variation in Infection State Uncertainty\u0026quot; https://t.co/ZvNghUHUUY #HMM #multievent #TMB #WomenInSTEM #carnivorepapersbywomen #heynas #serengeti pic.twitter.com/VT6iStk5UW\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 28 août 2018 ","date":1535414400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1535414400,"objectID":"e8bddb7f4ad792cd14721a438c39f725","permalink":"https://oliviergimenez.github.io/blog/sarahlulu_paper/","publishdate":"2018-08-28T00:00:00Z","relpermalink":"/blog/sarahlulu_paper/","section":"blog","summary":"New paper by #SarahBenhaiem \u0026amp; @LMarescot 🤩👏 \u0026quot;Robustness of Eco-Epidemiological Capture-Recapture Parameter Estimates to Variation in Infection State Uncertainty\u0026quot; https://t.co/ZvNghUHUUY #HMM #multievent #TMB #WomenInSTEM #carnivorepapersbywomen #heynas #serengeti pic.twitter.com/VT6iStk5UW\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 28 août 2018 ","tags":["multievent","capture-recapture","TMB","disease ecology","carnivores"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"","date":1534032000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1534032000,"objectID":"945017ff221838829cc693a7918149ae","permalink":"https://oliviergimenez.github.io/my-project/external-project/sciencepolicy/","publishdate":"2018-08-12T00:00:00Z","relpermalink":"/my-project/external-project/sciencepolicy/","section":"my-project","summary":"Combining ecological, human and social sciences for large carnivores management.","tags":["carnivores","ecology","social sciences","policy"],"title":"Socio-ecological management of large carnivores","type":"my-project"},{"authors":null,"categories":null,"content":"\u0026quot;Tenir compte de la dimension humaine dans la gestion des #conflits hommes-grands #carnivores : le cas de l’#ours brun dans les #Pyrenees\u0026quot; https://t.co/OJYC7Lje9L paru dans Faune Sauvage @ONCFS_officiel reprenant nos résultats publiés dans Biol. Cons. https://t.co/UgPsN37NSA pic.twitter.com/oRU7kLyd7Y\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 1 août 2018 ","date":1532217600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1532217600,"objectID":"70326c31b08f1284838e43542839e5ac","permalink":"https://oliviergimenez.github.io/blog/ours_faunesauvage/","publishdate":"2018-07-22T00:00:00Z","relpermalink":"/blog/ours_faunesauvage/","section":"blog","summary":"\u0026quot;Tenir compte de la dimension humaine dans la gestion des #conflits hommes-grands #carnivores : le cas de l’#ours brun dans les #Pyrenees\u0026quot; https://t.co/OJYC7Lje9L paru dans Faune Sauvage @ONCFS_officiel reprenant nos résultats publiés dans Biol.","tags":["scicomm","bear","social sciences"],"title":"Sciences sociales et ours brun dans les Pyrénées","type":"blog"},{"authors":null,"categories":null,"content":"🤩🐱 New paper by unstoppable @LouvrierJulie \u0026quot;Use of ambiguous detections to improve estimates from #SDMs\u0026quot; in @ConBiology https://t.co/XXZFGDT7ko #teamboulet @ZiFridolin @aveamphibious #WomenInSTEM #carnivorepapersbywomen @ONCFS_officiel @umontpellier @IsiteMUSE @CNRSenLR pic.twitter.com/LkZtbswikJ\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 20 juillet8 ","date":1532044800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1532044800,"objectID":"c706981fd52e0822c866f41be0f58617","permalink":"https://oliviergimenez.github.io/blog/julie_paper/","publishdate":"2018-07-20T00:00:00Z","relpermalink":"/blog/julie_paper/","section":"blog","summary":"🤩🐱 New paper by unstoppable @LouvrierJulie \u0026quot;Use of ambiguous detections to improve estimates from #SDMs\u0026quot; in @ConBiology https://t.co/XXZFGDT7ko #teamboulet @ZiFridolin @aveamphibious #WomenInSTEM #carnivorepapersbywomen @ONCFS_officiel @umontpellier @IsiteMUSE @CNRSenLR pic.twitter.com/LkZtbswikJ\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 20 juillet8 ","tags":["lynx","occupancy","misendification"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"Following @JamesGrecian\u0026#39;s tracks (who helped me to get the tweets), I did some analyses of the #isec2018 tweets largely inspired (if not copy and pasted) by previous analyses from @brenborbon, @_ColinFay and @bryangoodrich - Data \u0026amp; #rstats codes https://t.co/ZCv4DAeS5S pic.twitter.com/YmU2QBsyp2\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 9 juillt 2018 ","date":1531094400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1531094400,"objectID":"cc0ff6f61ddd2442474580f25fadc1aa","permalink":"https://oliviergimenez.github.io/blog/tweetisec2018/","publishdate":"2018-07-09T00:00:00Z","relpermalink":"/blog/tweetisec2018/","section":"blog","summary":"Following @JamesGrecian\u0026#39;s tracks (who helped me to get the tweets), I did some analyses of the #isec2018 tweets largely inspired (if not copy and pasted) by previous analyses from @brenborbon, @_ColinFay and @bryangoodrich - Data \u0026amp; #rstats codes https://t.","tags":["isec","statistical ecology","twitter"],"title":"ISEC 2018 tweets analysis","type":"blog"},{"authors":null,"categories":null,"content":"And nooooooow, ladies and gentlemen, @LMarescot presents her work on multispecies occupancy models to infer poaching #ISEC2018 @CNRSenLR @INEE_CNRS #teamboulet @WWF @SMARTCnsvTools pic.twitter.com/FUiATuATNJ\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 5et 2018 ","date":1530748800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1530748800,"objectID":"0bf67949b78078c93ecb253793920e65","permalink":"https://oliviergimenez.github.io/blog/isec2018lm/","publishdate":"2018-07-05T00:00:00Z","relpermalink":"/blog/isec2018lm/","section":"blog","summary":"And nooooooow, ladies and gentlemen, @LMarescot presents her work on multispecies occupancy models to infer poaching #ISEC2018 @CNRSenLR @INEE_CNRS #teamboulet @WWF @SMARTCnsvTools pic.twitter.com/FUiATuATNJ\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 5et 2018 ","tags":["isec","statistical ecology"],"title":"International Statistical Ecology Conference 2018 (4/4) - Lucile Marescot","type":"blog"},{"authors":null,"categories":null,"content":"Aaaaaaaaaaaaaaaaaaaand now, ladies and gentlemen, @NSantostasi on stage to present her work on estimating prevalence of hybrids in 🇮🇹 wolves using #HMMs and capture-recapture data #teamboulet @CNRSenLR @INEE_CNRS @romolocaniglia @umontpellier #CEFE pic.twitter.com/vDXH9NDX2H\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 5 juillet 2018 ","date":1530662400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1530662400,"objectID":"49c1a21b9db564e6e3ebdb1c55f8b9c5","permalink":"https://oliviergimenez.github.io/blog/isec2018ns/","publishdate":"2018-07-04T00:00:00Z","relpermalink":"/blog/isec2018ns/","section":"blog","summary":"Aaaaaaaaaaaaaaaaaaaand now, ladies and gentlemen, @NSantostasi on stage to present her work on estimating prevalence of hybrids in 🇮🇹 wolves using #HMMs and capture-recapture data #teamboulet @CNRSenLR @INEE_CNRS @romolocaniglia @umontpellier #CEFE pic.","tags":["isec","statistical ecology"],"title":"International Statistical Ecology Conference 2018 (3/4) - Nina Santostasi","type":"blog"},{"authors":null,"categories":null,"content":"Are you ready for an awesome talk by @oaggimenez on regularization techniques for covariate selection in #caprecapt studies with Ian Renner @MoonbeamLevels ? #ISEC2018 @ISECStatEcol 😍💪🎉 pic.twitter.com/XD4IRPgoL8\n\u0026mdash; Julie Louvrier 🐱🐺 (@LouvrierJulie) 3 juillet 2018 ","date":1530576000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1530576000,"objectID":"7852a339665e4409e8269024d7022c9b","permalink":"https://oliviergimenez.github.io/blog/isec2018og/","publishdate":"2018-07-03T00:00:00Z","relpermalink":"/blog/isec2018og/","section":"blog","summary":"Are you ready for an awesome talk by @oaggimenez on regularization techniques for covariate selection in #caprecapt studies with Ian Renner @MoonbeamLevels ? #ISEC2018 @ISECStatEcol 😍💪🎉 pic.twitter.com/XD4IRPgoL8\n\u0026mdash; Julie Louvrier 🐱🐺 (@LouvrierJulie) 3 juillet 2018 ","tags":["isec","statistical ecology"],"title":"International Statistical Ecology Conference 2018 (2/4) - Olivier Gimenez","type":"blog"},{"authors":null,"categories":null,"content":"And now, ladies and gentlemen, the amazing @LouvrierJulie 👏 who tells us about mechanistic #SDMs to study 🐺 colonization in 🇫🇷 #ISEC2018 #teamboulet @ONCFS_officiel @CNRSenLR @INEE_CNRS @umontpellier pic.twitter.com/JUTayPLTSu\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 2 juillet 2018 This is the end... pic.twitter.com/2FNCT8yxib\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 2 juillet 2018 ","date":1530489600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1530489600,"objectID":"71a794473f6ae36080f22eb670aed86a","permalink":"https://oliviergimenez.github.io/blog/isec2018jl/","publishdate":"2018-07-02T00:00:00Z","relpermalink":"/blog/isec2018jl/","section":"blog","summary":"And now, ladies and gentlemen, the amazing @LouvrierJulie 👏 who tells us about mechanistic #SDMs to study 🐺 colonization in 🇫🇷 #ISEC2018 #teamboulet @ONCFS_officiel @CNRSenLR @INEE_CNRS @umontpellier pic.twitter.com/JUTayPLTSu\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 2 juillet 2018 This is the end.","tags":["isec","statistical ecology"],"title":"International Statistical Ecology Conference 2018 (1/4) - Julie Louvrier","type":"blog"},{"authors":null,"categories":null,"content":"Ma présentation aux Rencontres Humains \u0026amp; Cétacés en Méditerranée https://t.co/rkEhfC7TPu du 27 Juin 👧👦🤝🐋🐬 à #Marseille sur l\u0026#39;apport de la recherche académique https://t.co/MEZtmisGPJ @GIS_3M #teamboulet #CEFE @CNRSenLR @INEE_CNRS\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 2uin 2018 ","date":1529971200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1529971200,"objectID":"fc74929cce60214dedac1d9cfe477ed5","permalink":"https://oliviergimenez.github.io/blog/rhc/","publishdate":"2018-06-26T00:00:00Z","relpermalink":"/blog/rhc/","section":"blog","summary":"Ma présentation aux Rencontres Humains \u0026amp; Cétacés en Méditerranée https://t.co/rkEhfC7TPu du 27 Juin 👧👦🤝🐋🐬 à #Marseille sur l\u0026#39;apport de la recherche académique https://t.co/MEZtmisGPJ @GIS_3M #teamboulet #CEFE @CNRSenLR @INEE_CNRS\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 2uin 2018 ","tags":["conference","cetaceans","scicomm"],"title":"Rencontres Humains \u0026 Cétacés en Méditerranée","type":"blog"},{"authors":null,"categories":null,"content":"What future for 🐘 in Laos: adaptation or transformation. pic.twitter.com/RIbcPg4svz\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 19 juin 2018 ","date":1529366400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1529366400,"objectID":"91a66ba55b450c3fb1e125dac0151e63","permalink":"https://oliviergimenez.github.io/blog/gilles_maurer_phdefense/","publishdate":"2018-06-19T00:00:00Z","relpermalink":"/blog/gilles_maurer_phdefense/","section":"blog","summary":"What future for 🐘 in Laos: adaptation or transformation. pic.twitter.com/RIbcPg4svz\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 19 juin 2018 ","tags":["phd","gender balance"],"title":"Gilles Maurer's PhD defense on the interactions between captive \u0026 wild pops of 🐘 w/ an interdisciplinary approach","type":"blog"},{"authors":null,"categories":null,"content":"We\u0026#39;re lucky to have Anne Loison on June 18 at #CEFE who\u0026#39;ll tell us about the determinants of gender gaps in career trajectories #WomenInSTEM @INEE_CNRS @CNRSenLR @CNRS @FemmesCNRS #EgaliteFH @FemmesSciences @umontpellier @IsiteMUSE pic.twitter.com/u82uloqoIJ\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 19 mai 20\u003e Throught provoking seminar by Anne Loison @Univ_Savoie #LECA on \u0026quot;Professional and personal trajectories of 👩 🔬👨 🔬: ♂️♀️ differences among ecologists in 🇫🇷 and 🇳🇴\u0026quot; https://t.co/8adLzUGCPd w/ paper here https://t.co/6lqX0vxeiQ @CNRSenLR @INEE_CNRS #CEFE #teamboulet pic.twitter.com/wMaYPDnvll\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 18 juin 2018 I liked Anne\u0026#39;s idea of role models for men very much, and the quote from Henry David Thoreau: pic.twitter.com/Ng0ffaV72n\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 18 juin 2018 ","date":1529280000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1529280000,"objectID":"113f6cf9fd64501c1651570a7682058e","permalink":"https://oliviergimenez.github.io/blog/anne_loison/","publishdate":"2018-06-18T00:00:00Z","relpermalink":"/blog/anne_loison/","section":"blog","summary":"We\u0026#39;re lucky to have Anne Loison on June 18 at #CEFE who\u0026#39;ll tell us about the determinants of gender gaps in career trajectories #WomenInSTEM @INEE_CNRS @CNRSenLR @CNRS @FemmesCNRS #EgaliteFH @FemmesSciences @umontpellier @IsiteMUSE pic.twitter.com/u82uloqoIJ\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 19 mai 20\u003e ","tags":["conference","gender balance"],"title":"Determinants of gender gaps in science","type":"blog"},{"authors":null,"categories":null,"content":"Linking organismal functions, life history strategies and population performance @BritishEcolSoc cross-journal special feature by the flamboyant @DRobcito w/ @dylan_childs #CyrilleViolle \u0026amp; I https://t.co/qoNviB84ux w/ contribs by @HalCaswell @djm_MEEG @benjaminblonder \u0026amp; others 🤩 pic.twitter.com/fqd6FvBrPz\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 18 juin8 ","date":1529280000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1529280000,"objectID":"a87d4d94d7f22b4c0c27bdcc184c1d14","permalink":"https://oliviergimenez.github.io/blog/bibi3_paper/","publishdate":"2018-06-18T00:00:00Z","relpermalink":"/blog/bibi3_paper/","section":"blog","summary":"Linking organismal functions, life history strategies and population performance @BritishEcolSoc cross-journal special feature by the flamboyant @DRobcito w/ @dylan_childs #CyrilleViolle \u0026amp; I https://t.co/qoNviB84ux w/ contribs by @HalCaswell @djm_MEEG @benjaminblonder \u0026amp; others 🤩 pic.","tags":["traits","demography"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"Au siège du @CNRS pour présenter notre projet combinant #shs et #écologie pour la gestion des grands prédateurs 🐺🐻🐱 en 🇫🇷 \u0026quot;Osez l\u0026#39;#interdisciplinarité\u0026quot; 🤞😓 @CNRSenLR @INEE_CNRS pic.twitter.com/yxUFWj5cWB\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 11 juin 2018 Et la réponse vient de tomber, ça maaaaaaaaaaaarche! 💥🎉😎🥳 pic.twitter.com/g8lI7l1kJW\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 15 juin 2018 ","date":1528675200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1528675200,"objectID":"8854b59ecd2923740cd5a049149a31dc","permalink":"https://oliviergimenez.github.io/blog/ecosocar/","publishdate":"2018-06-11T00:00:00Z","relpermalink":"/blog/ecosocar/","section":"blog","summary":"Au siège du @CNRS pour présenter notre projet combinant #shs et #écologie pour la gestion des grands prédateurs 🐺🐻🐱 en 🇫🇷 \u0026quot;Osez l\u0026#39;#interdisciplinarité\u0026quot; 🤞😓 @CNRSenLR @INEE_CNRS pic.twitter.com/yxUFWj5cWB\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 11 juin 2018 Et la réponse vient de tomber, ça maaaaaaaaaaaarche!","tags":["interdisciplinarity","carnivores","social sciences","species distribution"],"title":"Socio-ecological approaches to better understand interactions between large carnivores and human activities","type":"blog"},{"authors":null,"categories":null,"content":"Building w/ stakeholders an interface to assess lynx viability in 🇫🇷 Sarah Bauduin gives a crash course on spatially-explicit individual-based models (aka PacMan) #rstats #shiny #leaflet https://t.co/v9e865t2ib #teamboulet pic.twitter.com/1oTtRQRG1j\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 6 juin 2018 ","date":1528243200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1528243200,"objectID":"cfa9080fa8eb201103633899dd79b9a9","permalink":"https://oliviergimenez.github.io/blog/ittecop_meeting2/","publishdate":"2018-06-06T00:00:00Z","relpermalink":"/blog/ittecop_meeting2/","section":"blog","summary":"Building w/ stakeholders an interface to assess lynx viability in 🇫🇷 Sarah Bauduin gives a crash course on spatially-explicit individual-based models (aka PacMan) #rstats #shiny #leaflet https://t.co/v9e865t2ib #teamboulet pic.twitter.com/1oTtRQRG1j\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 6 juin 2018 ","tags":["interdisciplinarity","lynx","wildife-vehicle collisions"],"title":"Workshop on mitigating lynx-vehicle collisions","type":"blog"},{"authors":null,"categories":null,"content":"Special feature on Evolutionary #Demography by Takenori Takada \u0026amp; @Richshefferson in #Population #Ecology @SpringerJP_LS - We wrote \u0026quot;Estimating individual fitness in the wild using capture-recapture data\u0026quot; https://t.co/5u7x1UvRr5 #rstats code ➡️ https://t.co/l8o7Igydn9 pic.twitter.com/IUp6egBPUA\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 3 juin8 ","date":1527984000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1527984000,"objectID":"45852860f1c3554d60513c31bc0d9c96","permalink":"https://oliviergimenez.github.io/blog/bibi_paper2/","publishdate":"2018-06-03T00:00:00Z","relpermalink":"/blog/bibi_paper2/","section":"blog","summary":"Special feature on Evolutionary #Demography by Takenori Takada \u0026amp; @Richshefferson in #Population #Ecology @SpringerJP_LS - We wrote \u0026quot;Estimating individual fitness in the wild using capture-recapture data\u0026quot; https://t.co/5u7x1UvRr5 #rstats code ➡️ https://t.","tags":["fitness","capture-recapture"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"Aux journées de la @Fondationfrance @univpaulvalery \u0026quot;La mer monte, quel littoral pour demain\u0026quot; à la belle ville d\u0026#39;#Agde pour présenter notre programme #Interact 🐬⛵🎣 avec @GIS_3M https://t.co/72YF9gxCeE pic.twitter.com/HbRf40p6PX\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 31 mai 2018ockquote\u003e ","date":1527724800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1527724800,"objectID":"73b42cdc8c0846a0d9bae6511903538d","permalink":"https://oliviergimenez.github.io/blog/interact_media2/","publishdate":"2018-05-31T00:00:00Z","relpermalink":"/blog/interact_media2/","section":"blog","summary":"Aux journées de la @Fondationfrance @univpaulvalery \u0026quot;La mer monte, quel littoral pour demain\u0026quot; à la belle ville d\u0026#39;#Agde pour présenter notre programme #Interact 🐬⛵🎣 avec @GIS_3M https://t.co/72YF9gxCeE pic.twitter.com/HbRf40p6PX \u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 31 mai 2018ockquote\u003e ","tags":["conference","interact","dolphin"],"title":"La mer monte, quel littoral pour demain","type":"blog"},{"authors":null,"categories":null,"content":"Thrilled to have @pardo_deborah for our #teamboulet 🤩👩 🔬 Freelance scientist and coordinator @PureOceanFund @HomewardBound16 #WomenInSTEM #womeninscience #worklife balance #Marseille #networking #scicomm #howtohavefivejobsatonce #outreach pic.twitter.com/t8OX4miuDB\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 29 mai 2018","date":1527552000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1527552000,"objectID":"09d0a5fab06148c007a012d4e602d76f","permalink":"https://oliviergimenez.github.io/blog/deb_pardo/","publishdate":"2018-05-29T00:00:00Z","relpermalink":"/blog/deb_pardo/","section":"blog","summary":"Thrilled to have @pardo_deborah for our #teamboulet 🤩👩 🔬 Freelance scientist and coordinator @PureOceanFund @HomewardBound16 #WomenInSTEM #womeninscience #worklife balance #Marseille #networking #scicomm #howtohavefivejobsatonce #outreach pic.twitter.com/t8OX4miuDB\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 29 mai 2018","tags":["conference","scicomm"],"title":"Team meeting on communicating science and outreach","type":"blog"},{"authors":null,"categories":null,"content":"Had a stimulating @AgenceRecherche meeting working on #multispecies #demographic models to study structure \u0026amp; dynamics of 🐺🐦🐟🌱 #communities Check out https://t.co/eNFOzrPgWR w/ @UMR_MECADEV @isemevol #CESCO #CEFE #ONCFS @MARBEC_MicMac #CEBC @CNRSenLR @INEE_CNRS #teamboulet pic.twitter.com/XcfVCBWq8L\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 24 mai18 ","date":1527120000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1527120000,"objectID":"4ecb4432c6a4dfa095a2f4b140b48815","permalink":"https://oliviergimenez.github.io/blog/anr_meeting2/","publishdate":"2018-05-24T00:00:00Z","relpermalink":"/blog/anr_meeting2/","section":"blog","summary":"Had a stimulating @AgenceRecherche meeting working on #multispecies #demographic models to study structure \u0026amp; dynamics of 🐺🐦🐟🌱 #communities Check out https://t.co/eNFOzrPgWR w/ @UMR_MECADEV @isemevol #CESCO #CEFE #ONCFS @MARBEC_MicMac #CEBC @CNRSenLR @INEE_CNRS #teamboulet pic.","tags":["ANR","democom","multispecies"],"title":"Project meeting on multispecies demography","type":"blog"},{"authors":null,"categories":null,"content":"We had a stimulating #teamboulet meeting this morning, w/ #decocom https://t.co/eNFOzrPgWR post-doc #LucileMarescot giving a crash course on how to make optimal decisions under #uncertainty using stochastic dynamic programming #SDP #decisionmaking @AgenceRecherche @CNRSenLR pic.twitter.com/XkdSYjFGrL\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 22 mai 2018 ","date":1526947200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1526947200,"objectID":"1761a1a738c6b3c4f35f464570dd8251","permalink":"https://oliviergimenez.github.io/blog/boulet_meeting5/","publishdate":"2018-05-22T00:00:00Z","relpermalink":"/blog/boulet_meeting5/","section":"blog","summary":"We had a stimulating #teamboulet meeting this morning, w/ #decocom https://t.co/eNFOzrPgWR post-doc #LucileMarescot giving a crash course on how to make optimal decisions under #uncertainty using stochastic dynamic programming #SDP #decisionmaking @AgenceRecherche @CNRSenLR pic.","tags":["team"],"title":"Team meeting on stochastic dynamic programming","type":"blog"},{"authors":null,"categories":null,"content":"Awesome #teamboulet weekly meeting last Friday with #ReginaBestbier \u0026amp; @AmandaVincent1 @projectseahorse @UBC on communicating science 👩‍🔬👨‍🔬 \u0026quot;1/ Who is your audience 2/ What is your message 3/ Keep it simple and easy\u0026quot; #scicomm #WomenInSTEM @CNRSenLR pic.twitter.com/vl1x8X4IjM\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 21 mai 2018 ","date":1526860800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1526860800,"objectID":"5e1ca405910a542a511075e7bc374e3c","permalink":"https://oliviergimenez.github.io/blog/boulet_meeting4/","publishdate":"2018-05-21T00:00:00Z","relpermalink":"/blog/boulet_meeting4/","section":"blog","summary":"Awesome #teamboulet weekly meeting last Friday with #ReginaBestbier \u0026amp; @AmandaVincent1 @projectseahorse @UBC on communicating science 👩‍🔬👨‍🔬 \u0026quot;1/ Who is your audience 2/ What is your message 3/ Keep it simple and easy\u0026quot; #scicomm #WomenInSTEM @CNRSenLR pic.","tags":["team"],"title":"Team meeting on communicating science","type":"blog"},{"authors":null,"categories":null,"content":"Very honored and pleased 🤩 to serve as an Associate Editor for the new @BritishEcolSoc journal \u0026#39;People and Nature\u0026#39; 🎉🍾 Looking forward to handling my first submission 👀 #SocialSciences #ecology #nature #humanities https://t.co/2jl8FDdNqc\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 11 mai 2018 ","date":1525996800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1525996800,"objectID":"ca0cffb190a8816b7605adeae872843b","permalink":"https://oliviergimenez.github.io/blog/people_nature/","publishdate":"2018-05-11T00:00:00Z","relpermalink":"/blog/people_nature/","section":"blog","summary":"Very honored and pleased 🤩 to serve as an Associate Editor for the new @BritishEcolSoc journal \u0026#39;People and Nature\u0026#39; 🎉🍾 Looking forward to handling my first submission 👀 #SocialSciences #ecology #nature #humanities https://t.","tags":["People and Nature"],"title":"Associate editor for People and Nature","type":"blog"},{"authors":null,"categories":null,"content":"Séminaire interfaces #linguistique et #écologie organisé par @Praxiling et #CEFE 14/05 @montpellier_ On y parlera de notre article récent sur le traitement médiatique du 🐺 dans la 🗞️ https://t.co/D8R2tzt3D0 @CNRSenLR @univpaulvalery @lemondefr @lemonde_science @Nice_Matin pic.twitter.com/k8RfAfOzQF\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 9 mai 2018 ","date":1525824000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1525824000,"objectID":"d19137c368941954a286a054aba618a4","permalink":"https://oliviergimenez.github.io/blog/sem_praxiling/","publishdate":"2018-05-09T00:00:00Z","relpermalink":"/blog/sem_praxiling/","section":"blog","summary":"Séminaire interfaces #linguistique et #écologie organisé par @Praxiling et #CEFE 14/05 @montpellier_ On y parlera de notre article récent sur le traitement médiatique du 🐺 dans la 🗞️ https://t.co/D8R2tzt3D0 @CNRSenLR @univpaulvalery @lemondefr @lemonde_science @Nice_Matin pic.","tags":["interdisciplinarity"],"title":"Séminaire interfaces linguistique et écologie","type":"blog"},{"authors":null,"categories":null,"content":"Science 🌉 policy: Inspiring meeting this morning w/ Sarah Fowler @saveourseas scientific adviser and @AmandaVincent1 \u0026quot;How scientists can influence policy? One person can make a difference\u0026quot; #teamboulet #CEFE @INEE_CNRS @CNRSenLR @umontpellier pic.twitter.com/S3J8jmz75d\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 7 mai 2018 ","date":1525651200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1525651200,"objectID":"cd0b67f64c48e4a00e9994f5ef452439","permalink":"https://oliviergimenez.github.io/blog/boulet_meeting3/","publishdate":"2018-05-07T00:00:00Z","relpermalink":"/blog/boulet_meeting3/","section":"blog","summary":"Science 🌉 policy: Inspiring meeting this morning w/ Sarah Fowler @saveourseas scientific adviser and @AmandaVincent1 \u0026quot;How scientists can influence policy? One person can make a difference\u0026quot; #teamboulet #CEFE @INEE_CNRS @CNRSenLR @umontpellier pic.","tags":["team"],"title":"Team meeting science and policy","type":"blog"},{"authors":null,"categories":null,"content":"\u0026quot;R2ucare: An R #rstats package to perform goodness‐of‐fit tests for capture‐recapture models\u0026quot; in @MethodsEcolEvol https://t.co/ieQohrn3jA 🤓😎 - paper entirely written w/ #rmarkdown thx @stevenvmiller https://t.co/9FapGe3oGV 💯 - replication files ➡️ https://t.co/h9F6sKbcpI pic.twitter.com/JyzQ33Wrl3\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 21 avril 2018 ","date":1524268800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1524268800,"objectID":"374ffda78aad73d73e4744f1928d4b55","permalink":"https://oliviergimenez.github.io/blog/bibi_paper/","publishdate":"2018-04-21T00:00:00Z","relpermalink":"/blog/bibi_paper/","section":"blog","summary":"\u0026quot;R2ucare: An R #rstats package to perform goodness‐of‐fit tests for capture‐recapture models\u0026quot; in @MethodsEcolEvol https://t.co/ieQohrn3jA 🤓😎 - paper entirely written w/ #rmarkdown thx @stevenvmiller https://t.co/9FapGe3oGV 💯 - replication files ➡️ https://t.","tags":["gof","capture-recapture"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"Ma thèse en 180 secondes (2018) - Des loups et des crottes 🐺💩 - Julie Louvrier https://t.co/DZMwANiEYD ou comment vulgariser son travail de #thèse sur la modélisation #statistique de la distribution des grands #carnivores #occupancy #SDM #MT180\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 16 avril 2018 ","date":1523836800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1523836800,"objectID":"cbfc67f8c887642766b09cec8c1df87b","permalink":"https://oliviergimenez.github.io/blog/julie_mt180/","publishdate":"2018-04-16T00:00:00Z","relpermalink":"/blog/julie_mt180/","section":"blog","summary":"Ma thèse en 180 secondes (2018) - Des loups et des crottes 🐺💩 - Julie Louvrier https://t.co/DZMwANiEYD ou comment vulgariser son travail de #thèse sur la modélisation #statistique de la distribution des grands #carnivores #occupancy #SDM #MT180","tags":["scicomm","wolf"],"title":"Julie dans MT180s","type":"blog"},{"authors":null,"categories":null,"content":"A career path from #academia to #datascience - inspiring #teamboulet meeting today w/ @StatnMap from @thinkR_fr. I\u0026#39;m sharing a few #careertips he gave us 💯 👇\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 3 avril 2018 ","date":1522800000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1522800000,"objectID":"cfbc006daf09f7bc07c7ceb8205be12b","permalink":"https://oliviergimenez.github.io/blog/team_boulet2/","publishdate":"2018-04-04T00:00:00Z","relpermalink":"/blog/team_boulet2/","section":"blog","summary":"A career path from #academia to #datascience - inspiring #teamboulet meeting today w/ @StatnMap from @thinkR_fr. I\u0026#39;m sharing a few #careertips he gave us 💯 👇\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 3 avril 2018 ","tags":["team","data science"],"title":"Team meeting: A career path from academia to data science","type":"blog"},{"authors":null,"categories":null,"content":"Awesome day of talks/discussions between the 🇫🇷 Game and Wildlife Agency #ONCFS and @INEE_CNRS to reinforce our links and improve the ways we work together @NirSeMa 🐹🐰🐻🐸🦊🐺🐦🐗🦌🦆 pic.twitter.com/DePhj2Qfnw\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 27 mars 2018 ","date":1522108800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1522108800,"objectID":"bbff32cac82a6c8686b681b5afc7ca2c","permalink":"https://oliviergimenez.github.io/blog/cnrs_oncfs_day/","publishdate":"2018-03-27T00:00:00Z","relpermalink":"/blog/cnrs_oncfs_day/","section":"blog","summary":"Awesome day of talks/discussions between the 🇫🇷 Game and Wildlife Agency #ONCFS and @INEE_CNRS to reinforce our links and improve the ways we work together @NirSeMa 🐹🐰🐻🐸🦊🐺🐦🐗🦌🦆 pic.twitter.com/DePhj2Qfnw\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 27 mars 2018 ","tags":["oncfs","wildlife management"],"title":"Journées d'échanges entre ONCFS et CNRS","type":"blog"},{"authors":null,"categories":null,"content":"Présentation de notre projet Interact https://t.co/72YF9gxCeE sur interactions dauphins-activités humaines dans le bulletin Pelagos 2018 https://t.co/rVHPhz1pnt @Fondationfrance @GIS_3M @Pelagos_ @INEE_CNRS @CNRSenLR pic.twitter.com/RYRlcxmYAq\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 20 mars 2018 ","date":1521590400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1521590400,"objectID":"ef058c2f3d0c0cefdc56c93413a286dc","permalink":"https://oliviergimenez.github.io/blog/interact_media/","publishdate":"2018-03-21T00:00:00Z","relpermalink":"/blog/interact_media/","section":"blog","summary":"Présentation de notre projet Interact https://t.co/72YF9gxCeE sur interactions dauphins-activités humaines dans le bulletin Pelagos 2018 https://t.co/rVHPhz1pnt @Fondationfrance @GIS_3M @Pelagos_ @INEE_CNRS @CNRSenLR pic.twitter.com/RYRlcxmYAq\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 20 mars 2018 ","tags":["interact","cetaceans","interdisciplinarity"],"title":"Interactions between dolphins and human activities","type":"blog"},{"authors":null,"categories":null,"content":"#multievent #capturerecapture workshop - Intro by #RogerPradel 😍📈 Check out the comments for a few relevant papers on the topic @INEE_CNRS @CNRSenLR pic.twitter.com/4nHrXBakgf\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 20 mars 2018 ","date":1521504000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1521504000,"objectID":"5818b15a0a7342ff5d03303b6dde2fcb","permalink":"https://oliviergimenez.github.io/blog/capturerecapture_workshop2018/","publishdate":"2018-03-20T00:00:00Z","relpermalink":"/blog/capturerecapture_workshop2018/","section":"blog","summary":"#multievent #capturerecapture workshop - Intro by #RogerPradel 😍📈 Check out the comments for a few relevant papers on the topic @INEE_CNRS @CNRSenLR pic.twitter.com/4nHrXBakgf\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 20 mars 2018 ","tags":["team","workshop","capture-recapture"],"title":"2018 capture-recapture workshop","type":"blog"},{"authors":null,"categories":null,"content":"Inspiring #teamboulet meeting on the role of scientists in public debate with special guest sociologist #LauraSeguin https://t.co/4DQyDYJ0vA #WomenInScience @FredBarraquand @AmandaVincent1 @ValentinLauret @SCubi25 pic.twitter.com/T4eY8hiN6H\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 17 mars 2018 ","date":1521244800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1521244800,"objectID":"33d7ef34f0ab046d02783c3a900ca057","permalink":"https://oliviergimenez.github.io/blog/boulet_meeting/","publishdate":"2018-03-17T00:00:00Z","relpermalink":"/blog/boulet_meeting/","section":"blog","summary":"Inspiring #teamboulet meeting on the role of scientists in public debate with special guest sociologist #LauraSeguin https://t.co/4DQyDYJ0vA #WomenInScience @FredBarraquand @AmandaVincent1 @ValentinLauret @SCubi25 pic.twitter.com/T4eY8hiN6H\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 17 mars 2018 ","tags":["team"],"title":"Team meeting on the role of scientists in public debate","type":"blog"},{"authors":null,"categories":null,"content":"Thrilled to welcome @AmandaVincent1 for a 6-month sabbatical in the #teamboulet @projectseahorse https://t.co/dcpXbRDJHB @INEE_CNRS @CNRSenLR pic.twitter.com/OyMbMKwO9K\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 17 mars 2018 ","date":1520380800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1520380800,"objectID":"8fc947f45f755245b2380f9f847016b0","permalink":"https://oliviergimenez.github.io/blog/amanda_sabbatical/","publishdate":"2018-03-07T00:00:00Z","relpermalink":"/blog/amanda_sabbatical/","section":"blog","summary":"Thrilled to welcome @AmandaVincent1 for a 6-month sabbatical in the #teamboulet @projectseahorse https://t.co/dcpXbRDJHB @INEE_CNRS @CNRSenLR pic.twitter.com/OyMbMKwO9K\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 17 mars 2018 ","tags":["interdisciplinarity","marine realm","conservation"],"title":"Amanda Vincent is on sabbatical with us","type":"blog"},{"authors":null,"categories":null,"content":"New paper 🎉🎉 \u0026quot;Social status mediates the fitness costs of infection with canine distemper virus in Serengeti spotted hyenas\u0026quot; https://t.co/LdWNAWFWum @BritishEcolSoc by L. Marescot and S. Benhaiem #teamboulet #multievent #hmm 😎😊 https://t.co/JKpN1scRTx pic.twitter.com/pD7yHGnhZ5\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 7 mars 2018 ","date":1520380800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1520380800,"objectID":"4d343cd3c78b6ce3da0959358cf820da","permalink":"https://oliviergimenez.github.io/blog/lulu_paper/","publishdate":"2018-03-07T00:00:00Z","relpermalink":"/blog/lulu_paper/","section":"blog","summary":"New paper 🎉🎉 \u0026quot;Social status mediates the fitness costs of infection with canine distemper virus in Serengeti spotted hyenas\u0026quot; https://t.co/LdWNAWFWum @BritishEcolSoc by L. Marescot and S. Benhaiem #teamboulet #multievent #hmm 😎😊 https://t.","tags":["interdisciplinarity","hyenas","capture-recapture"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"Journée prospective de notre laboratoire le #CEFE @INEE_CNRS @CNRSenLR @umontpellier 😍😉 pic.twitter.com/WwUDndaEgO\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 6 mars 2018 ","date":1520294400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1520294400,"objectID":"1ba968957849075938ddd9cb0fd36c58","permalink":"https://oliviergimenez.github.io/blog/prospective_cefe/","publishdate":"2018-03-06T00:00:00Z","relpermalink":"/blog/prospective_cefe/","section":"blog","summary":"Journée prospective de notre laboratoire le #CEFE @INEE_CNRS @CNRSenLR @umontpellier 😍😉 pic.twitter.com/WwUDndaEgO\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 6 mars 2018 ","tags":["CEFE"],"title":"Prospectives scientifiques du CEFE","type":"blog"},{"authors":null,"categories":null,"content":"Réunion de lancement avec les acteurs du territoire de notre projet sur \u0026quot;Eviter, réduire et compenser la mortalité du Lynx par collision avec les véhicules de transport\u0026quot; https://t.co/uKY5ylOli2 @_Cerema #oncfs @CROC_Estelle @INEE_CNRS #ittecop @ademe @FRBiodiv @Min_Ecologie 🤩💪 pic.twitter.com/3sY0rDIODQ\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 5 mars 2018 ","date":1520208000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1520208000,"objectID":"3b3f6c8cb1771f0100eddd62636ac33e","permalink":"https://oliviergimenez.github.io/blog/ittecop_meeting/","publishdate":"2018-03-05T00:00:00Z","relpermalink":"/blog/ittecop_meeting/","section":"blog","summary":"Réunion de lancement avec les acteurs du territoire de notre projet sur \u0026quot;Eviter, réduire et compenser la mortalité du Lynx par collision avec les véhicules de transport\u0026quot; https://t.co/uKY5ylOli2 @_Cerema #oncfs @CROC_Estelle @INEE_CNRS #ittecop @ademe @FRBiodiv @Min_Ecologie 🤩💪 pic.","tags":["interdisciplinarity","lynx","wildife-vehicle collisions"],"title":"Kick-off meeting of our project on mitigating lynx-vehicle collisions","type":"blog"},{"authors":null,"categories":null,"content":"New paper \u0026quot;Searching for meaning in marine mammal shared data\u0026quot; 🐬🖥️📈 #openaccess #reproducibility https://t.co/puZrXHACUW pic.twitter.com/7ULeAJLMfT\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 23 février 2018 ","date":1519344000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1519344000,"objectID":"2fd897cadb05e896ceb3c412d15ff043","permalink":"https://oliviergimenez.github.io/blog/gio_paper/","publishdate":"2018-02-23T00:00:00Z","relpermalink":"/blog/gio_paper/","section":"blog","summary":"New paper \u0026quot;Searching for meaning in marine mammal shared data\u0026quot; 🐬🖥️📈 #openaccess #reproducibility https://t.co/puZrXHACUW pic.twitter.com/7ULeAJLMfT\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 23 février 2018 ","tags":["interdisciplinarity","cetaceans","data sharing"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"New paper on \u0026quot;Content analysis of 🗞️ coverage of 🐺 recolonization in 🇫🇷 using structural topic modeling\u0026quot;; free PDF here: https://t.co/w4wHPfbSZ8 (there are production errors in the figures that should be corrected in a few days) pic.twitter.com/o8Qo9nY7zC\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 23 février 2018 ","date":1519344000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1519344000,"objectID":"3cd411348165eff5d0f07644f7d142bd","permalink":"https://oliviergimenez.github.io/blog/marie_paper/","publishdate":"2018-02-23T00:00:00Z","relpermalink":"/blog/marie_paper/","section":"blog","summary":"New paper on \u0026quot;Content analysis of 🗞️ coverage of 🐺 recolonization in 🇫🇷 using structural topic modeling\u0026quot;; free PDF here: https://t.co/w4wHPfbSZ8 (there are production errors in the figures that should be corrected in a few days) pic.","tags":["interdisciplinarity","wolf","content analysis","media"],"title":"New paper!","type":"blog"},{"authors":null,"categories":null,"content":"Philosophie, linguistique, géographie, psychologie, écologie : la question de la nature. #SHS dans un labo d écologie et d évolution! #femmesenscience #WomenInScience pic.twitter.com/mEQcur0EKv\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 16 février 2018 ","date":1518739200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1518739200,"objectID":"270346b48ad9793cdf4613acd0717ed0","permalink":"https://oliviergimenez.github.io/blog/biocon_meeting/","publishdate":"2018-02-16T00:00:00Z","relpermalink":"/blog/biocon_meeting/","section":"blog","summary":"Philosophie, linguistique, géographie, psychologie, écologie : la question de la nature. #SHS dans un labo d écologie et d évolution! #femmesenscience #WomenInScience pic.twitter.com/mEQcur0EKv\n\u0026mdash; Olivier Gimenez 🍉 (@oaggimenez) 16 février 2018 ","tags":["interdisciplinarity","conservation","CEFE"],"title":"Annual meeting our Biodiversity and Conservation meeting","type":"blog"},{"authors":null,"categories":null,"content":" Recently, I have been using `OpenBUGS` for some analyses that `JAGS` cannot do. However, `JAGS` can be run in parallel through [the `jagsUI` package](https://github.com/kenkellner/jagsUI), which can save you some precious time. So the question is how to run several chains in parallel with `OpenBUGS`. Well, first you\u0026rsquo;ll need to install OpenBUGS (if you\u0026rsquo;re on a Mac, check out this short tutorial). Then, you\u0026rsquo;ll need to run OpenBUGS from R through the pacage R2OpenBUGS, which you can install via:\nif(!require(R2OpenBUGS)) install.packages(\u0026quot;R2OpenBUGS\u0026quot;) ## Loading required package: R2OpenBUGS Standard analysis Now let\u0026rsquo;s run the classical BUGS school example:\nLoad the OpenBUGS Package\nlibrary(R2OpenBUGS) Load the data\ndata(schools) Define the model, write it to a text file and have a look\nnummodel \u0026lt;- function(){ for (j in 1:J){ y[j] ~ dnorm (theta[j], tau.y[j]) theta[j] ~ dnorm (mu.theta, tau.theta) tau.y[j] \u0026lt;- pow(sigma.y[j], -2)} mu.theta ~ dnorm (0.0, 1.0E-6) tau.theta \u0026lt;- pow(sigma.theta, -2) sigma.theta ~ dunif (0, 1000) } write.model(nummodel, \u0026quot;nummodel.txt\u0026quot;) model.file1 = paste(getwd(),\u0026quot;nummodel.txt\u0026quot;, sep=\u0026quot;/\u0026quot;) file.show(\u0026quot;nummodel.txt\u0026quot;) Prepare the data for input into OpenBUGS\nJ \u0026lt;- nrow(schools) y \u0026lt;- schools$estimate sigma.y \u0026lt;- schools$sd data \u0026lt;- list (\u0026quot;J\u0026quot;, \u0026quot;y\u0026quot;, \u0026quot;sigma.y\u0026quot;) Initialization of variables\ninits \u0026lt;- function(){ list(theta = rnorm(J, 0, 100), mu.theta = rnorm(1, 0, 100), sigma.theta = runif(1, 0, 100))} Set the Wine working directory and the directory to OpenBUGS, and change the OpenBUGS.exe location as necessary:\nWINE=\u0026quot;/usr/local/Cellar/wine/2.0.4/bin/wine\u0026quot; WINEPATH=\u0026quot;/usr/local/Cellar/wine/2.0.4/bin/winepath\u0026quot; OpenBUGS.pgm=\u0026quot;/Applications/OpenBUGS323/OpenBUGS.exe\u0026quot; The are the parameters to save\nparameters = c(\u0026quot;theta\u0026quot;, \u0026quot;mu.theta\u0026quot;, \u0026quot;sigma.theta\u0026quot;) Run the model\nptm \u0026lt;- proc.time() schools.sim \u0026lt;- bugs(data, inits, model.file = model.file1,parameters=parameters,n.chains = 2, n.iter = 500000, n.burnin = 10000, OpenBUGS.pgm=OpenBUGS.pgm, WINE=WINE, WINEPATH=WINEPATH,useWINE=T) elapsed_time \u0026lt;- proc.time() - ptm elapsed_time ## user system elapsed ## 50.835 2.053 55.010 print(schools.sim) ## Inference for Bugs model at \u0026quot;/Users/oliviergimenez/Desktop/nummodel.txt\u0026quot;, ## Current: 2 chains, each with 5e+05 iterations (first 10000 discarded) ## Cumulative: n.sims = 980000 iterations saved ## mean sd 2.5% 25% 50% 75% 97.5% Rhat n.eff ## theta[1] 11.6 8.4 -1.9 6.1 10.5 15.8 32.0 1 370000 ## theta[2] 8.0 6.4 -4.8 4.0 8.0 12.0 20.9 1 67000 ## theta[3] 6.4 7.8 -11.3 2.2 6.8 11.2 20.9 1 55000 ## theta[4] 7.7 6.6 -5.7 3.7 7.8 11.8 20.9 1 70000 ## theta[5] 5.5 6.5 -8.8 1.6 5.9 9.8 17.1 1 26000 ## theta[6] 6.2 6.9 -8.9 2.3 6.6 10.7 18.9 1 23000 ## theta[7] 10.7 6.9 -1.4 6.0 10.1 14.7 26.2 1 480000 ## theta[8] 8.7 7.9 -6.8 4.0 8.4 13.0 25.7 1 76000 ## mu.theta 8.1 5.3 -2.0 4.7 8.1 11.4 18.5 1 30000 ## sigma.theta 6.6 5.7 0.2 2.5 5.2 9.1 20.9 1 12000 ## deviance 60.5 2.2 57.0 59.1 60.1 61.4 66.0 1 980000 ## ## For each parameter, n.eff is a crude measure of effective sample size, ## and Rhat is the potential scale reduction factor (at convergence, Rhat=1). ## ## DIC info (using the rule, pD = Dbar-Dhat) ## pD = 2.8 and DIC = 63.2 ## DIC is an estimate of expected predictive error (lower deviance is better). Parallel computations To run several chains in parallel, we\u0026rsquo;ll follow the steps described in this nice post.\n# loading packages library(snow) library(snowfall) # setting the number of CPUs to be 2 sfInit(parallel=TRUE, cpus=2) ## Warning in searchCommandline(parallel, cpus = cpus, type ## = type, socketHosts = socketHosts, : Unknown option on ## commandline: rmarkdown::render('/Users/oliviergimenez/Desktop/ ## run_openbugs_in_parallel.Rmd',~+~~+~encoding~+~ ## R Version: R version 3.4.3 (2017-11-30) ## snowfall 1.84-6.1 initialized (using snow 0.4-2): parallel execution on 2 CPUs. # and assigning the R2OpenBUGS library to each CPU sfLibrary(R2OpenBUGS) ## Library R2OpenBUGS loaded. ## Library R2OpenBUGS loaded in cluster. # create list of data J \u0026lt;- nrow(schools) y \u0026lt;- schools$estimate sigma.y \u0026lt;- schools$sd x.data \u0026lt;- list (J=J, y=y, sigma.y=sigma.y) # creating separate directory for each CPU process folder1 \u0026lt;- paste(getwd(), \u0026quot;/chain1\u0026quot;, sep=\u0026quot;\u0026quot;) folder2 \u0026lt;- paste(getwd(), \u0026quot;/chain2\u0026quot;, sep=\u0026quot;\u0026quot;) dir.create(folder1); dir.create(folder2); # sinking the model into a file in each directory for (folder in c(folder1, folder2)) { sink(paste(folder, \u0026quot;/nummodel.txt\u0026quot;, sep=\u0026quot;\u0026quot;)) cat(\u0026quot; model{ for (j in 1:J){ y[j] ~ dnorm (theta[j], tau.y[j]) theta[j] ~ dnorm (mu.theta, tau.theta) tau.y[j] \u0026lt;- pow(sigma.y[j], -2)} mu.theta ~ dnorm (0.0, 1.0E-6) tau.theta \u0026lt;- pow(sigma.theta, -2) sigma.theta ~ dunif (0, 1000) } \u0026quot;) sink() } # defining the function that will run MCMC on each CPU # Arguments: # chain - will be 1 or 2 # x.data - the data list # params - parameters to be monitored parallel.bugs \u0026lt;- function(chain, x.data, params) { # a. defining directory for each CPU sub.folder \u0026lt;- paste(getwd(),\u0026quot;/chain\u0026quot;, chain, sep=\u0026quot;\u0026quot;) # b. specifying the initial MCMC values inits \u0026lt;- function()list(theta = rnorm(x.data$J, 0, 100), mu.theta = rnorm(1, 0, 100), sigma.theta = runif(1, 0, 100)) # c. calling OpenBugs # (you may need to change the OpenBUGS.pgm directory) # je suis sous Mac, je fais tourner OpenBUGS via Wine bugs(data=x.data, inits=inits, parameters.to.save=params, n.iter = 500000, n.burnin = 10000, n.chains=1, model.file=\u0026quot;nummodel.txt\u0026quot;, debug=FALSE, codaPkg=TRUE, useWINE=TRUE, OpenBUGS.pgm = \u0026quot;/Applications/OpenBUGS323/OpenBUGS.exe\u0026quot;, working.directory = sub.folder, WINE=\u0026quot;/usr/local/Cellar/wine/2.0.4/bin/wine\u0026quot;, WINEPATH=\u0026quot;/usr/local/Cellar/wine/2.0.4/bin/winepath\u0026quot;) } # setting the parameters to be monitored params \u0026lt;- c(\u0026quot;theta\u0026quot;, \u0026quot;mu.theta\u0026quot;, \u0026quot;sigma.theta\u0026quot;) # calling the sfLapply function that will run # parallel.bugs on each of the 2 CPUs ptm \u0026lt;- proc.time() sfLapply(1:2, fun=parallel.bugs, x.data=x.data, params=params) ## [[1]] ## [1] \u0026quot;/Users/oliviergimenez/Desktop/chain1/CODAchain1.txt\u0026quot; ## ## [[2]] ## [1] \u0026quot;/Users/oliviergimenez/Desktop/chain2/CODAchain1.txt\u0026quot; elapsed_time = proc.time() - ptm elapsed_time ## user system elapsed ## 0.013 0.000 32.157 # locating position of each CODA chain chain1 \u0026lt;- paste(folder1, \u0026quot;/CODAchain1.txt\u0026quot;, sep=\u0026quot;\u0026quot;) chain2 \u0026lt;- paste(folder2, \u0026quot;/CODAchain1.txt\u0026quot;, sep=\u0026quot;\u0026quot;) # and, finally, getting the results res \u0026lt;- read.bugs(c(chain1, chain2)) ## Abstracting deviance ... 490000 valid values ## Abstracting mu.theta ... 490000 valid values ## Abstracting sigma.theta ... 490000 valid values ## Abstracting theta[1] ... 490000 valid values ## Abstracting theta[2] ... 490000 valid values ## Abstracting theta[3] ... 490000 valid values ## Abstracting theta[4] ... 490000 valid values ## Abstracting theta[5] ... 490000 valid values ## Abstracting theta[6] ... 490000 valid values ## Abstracting theta[7] ... 490000 valid values ## Abstracting theta[8] ... 490000 valid values ## Abstracting deviance ... 490000 valid values ## Abstracting mu.theta ... 490000 valid values ## Abstracting sigma.theta ... 490000 valid values ## Abstracting theta[1] ... 490000 valid values ## Abstracting theta[2] ... 490000 valid values ## Abstracting theta[3] ... 490000 valid values ## Abstracting theta[4] ... 490000 valid values ## Abstracting theta[5] ... 490000 valid values ## Abstracting theta[6] ... 490000 valid values ## Abstracting theta[7] ... 490000 valid values ## Abstracting theta[8] ... 490000 valid values summary(res) ## ## Iterations = 10001:5e+05 ## Thinning interval = 1 ## Number of chains = 2 ## Sample size per chain = 490000 ## ## 1. Empirical mean and standard deviation for each variable, ## plus standard error of the mean: ## ## Mean SD Naive SE Time-series SE ## deviance 60.453 2.221 0.002243 0.005737 ## mu.theta 8.109 5.261 0.005315 0.020596 ## sigma.theta 6.610 5.682 0.005740 0.027336 ## theta[1] 11.697 8.407 0.008493 0.027974 ## theta[2] 8.023 6.395 0.006460 0.019264 ## theta[3] 6.365 7.866 0.007946 0.022485 ## theta[4] 7.735 6.601 0.006668 0.019766 ## theta[5] 5.467 6.504 0.006570 0.022580 ## theta[6] 6.234 6.885 0.006955 0.021332 ## theta[7] 10.727 6.891 0.006961 0.023304 ## theta[8] 8.648 7.892 0.007972 0.021541 ## ## 2. Quantiles for each variable: ## ## 2.5% 25% 50% 75% 97.5% ## deviance 57.0200 59.120 60.040 61.430 65.99 ## mu.theta -2.0600 4.784 8.066 11.410 18.50 ## sigma.theta 0.2275 2.456 5.275 9.190 20.82 ## theta[1] -1.8850 6.195 10.560 15.880 32.04 ## theta[2] -4.8350 4.049 8.004 11.980 20.96 ## theta[3] -11.4800 2.194 6.871 11.170 20.90 ## theta[4] -5.7500 3.711 7.784 11.820 20.92 ## theta[5] -8.8490 1.603 5.938 9.834 17.14 ## theta[6] -8.8940 2.255 6.632 10.680 18.95 ## theta[7] -1.3450 6.125 10.140 14.680 26.27 ## theta[8] -6.8910 4.037 8.409 12.960 25.72 ","date":1515888000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1515888000,"objectID":"b5b5873064251d05885f45fb42d7bb69","permalink":"https://oliviergimenez.github.io/blog/run_openbugs_parallel/","publishdate":"2018-01-14T00:00:00Z","relpermalink":"/blog/run_openbugs_parallel/","section":"blog","summary":" Recently, I have been using `OpenBUGS` for some analyses that `JAGS` cannot do. However, `JAGS` can be run in parallel through [the `jagsUI` package](https://github.com/kenkellner/jagsUI), which can save you some precious time. So the question is how to run several chains in parallel with `OpenBUGS`.","tags":["bugs","R","rstats","OpenBUGS","parallel computation"],"title":"Running OpenBUGS in parallel","type":"blog"},{"authors":null,"categories":null,"content":" I had to use the good old `OpenBUGS` for some analyses that cannot be done in `JAGS`. Below are the steps to install `OpenBUGS` then to run it from your Mac either natively or from `R`. This tutorial is an adaptation of [this post](https://sites.google.com/site/mmeclimate/-bayesmet/openbugs-on-mac-os-x) and [that one](http://www.davideagle.org/r-2/bayesian-modeling-using-winbugs-and-openbugs/running-openbugs-on-mac-using-wine). If not done already, install Homebrew. This program will make the installation of any other programs on your Mac so easy!\nInstall Wine which will allow you to run any Windows programs (.exe) on your Mac. To do so, start by opening Terminal, then type in the command: brew install wine\nNext, download the Windows version of OpenBUGS here\nTo install OpenBUGS, still in Terminal, go to the directory where the file was downloaded and type (you might need to unzip the file you downloaded first): wine OpenBUGS323setup.exe\nOpenBUGS is now installed and ready to be used! You can run it by first going to the directory where OpenBUGS was installed. On my laptop, it can be achieved via the command: cd /Applications/OpenBUGS323\nThen, you just need to tye in the following command in the Terminal, and you should see an OpenBUGS windows poping up: wine OpenBUGS\nNow we would like to run OpenBUGS from R.\nInstall the package R2OpenBUGS by typing in the R console: if(!require(R2OpenBUGS)) install.packages(\u0026quot;R2OpenBUGS\u0026quot;) ## Loading required package: R2OpenBUGS Now let\u0026rsquo;s see whether everything works well by running the classical BUGS school example: Load the OpenBUGS Package\nlibrary(R2OpenBUGS) Load the data\ndata(schools) Define the model, write it to a text file and have a look\nnummodel \u0026lt;- function(){ for (j in 1:J){ y[j] ~ dnorm (theta[j], tau.y[j]) theta[j] ~ dnorm (mu.theta, tau.theta) tau.y[j] \u0026lt;- pow(sigma.y[j], -2)} mu.theta ~ dnorm (0.0, 1.0E-6) tau.theta \u0026lt;- pow(sigma.theta, -2) sigma.theta ~ dunif (0, 1000) } write.model(nummodel, \u0026quot;nummodel.txt\u0026quot;) model.file1 = paste(getwd(),\u0026quot;nummodel.txt\u0026quot;, sep=\u0026quot;/\u0026quot;) file.show(\u0026quot;nummodel.txt\u0026quot;) Prepare the data for input into OpenBUGS\nJ \u0026lt;- nrow(schools) y \u0026lt;- schools$estimate sigma.y \u0026lt;- schools$sd data \u0026lt;- list (\u0026quot;J\u0026quot;, \u0026quot;y\u0026quot;, \u0026quot;sigma.y\u0026quot;) Initialization of variables\ninits \u0026lt;- function(){ list(theta = rnorm(J, 0, 100), mu.theta = rnorm(1, 0, 100), sigma.theta = runif(1, 0, 100))} Set the Wine working directory and the directory to OpenBUGS, and change the OpenBUGS.exe location as necessary:\nWINE=\u0026quot;/usr/local/Cellar/wine/2.0.4/bin/wine\u0026quot; WINEPATH=\u0026quot;/usr/local/Cellar/wine/2.0.4/bin/winepath\u0026quot; OpenBUGS.pgm=\u0026quot;/Applications/OpenBUGS323/OpenBUGS.exe\u0026quot; The are the parameters to save\nparameters = c(\u0026quot;theta\u0026quot;, \u0026quot;mu.theta\u0026quot;, \u0026quot;sigma.theta\u0026quot;) Run the model\nschools.sim \u0026lt;- bugs(data, inits, model.file = model.file1,parameters=parameters,n.chains = 3, n.iter = 1000, OpenBUGS.pgm=OpenBUGS.pgm, WINE=WINE, WINEPATH=WINEPATH,useWINE=T) R will pause. You might get a weird message starting by err:ole, just ignore it. When the run is complete, a prompt will reappear, then just type the following command to get the result:\nprint(schools.sim) ## Inference for Bugs model at \u0026quot;/Users/oliviergimenez/Desktop/nummodel.txt\u0026quot;, ## Current: 3 chains, each with 1000 iterations (first 500 discarded) ## Cumulative: n.sims = 1500 iterations saved ## mean sd 2.5% 25% 50% 75% 97.5% Rhat n.eff ## theta[1] 12.2 7.9 -1.3 7.5 11.2 16.4 32.1 1.0 62 ## theta[2] 9.1 6.5 -4.0 5.1 9.4 13.2 21.4 1.0 150 ## theta[3] 7.8 7.7 -9.4 3.6 8.5 12.6 21.1 1.0 360 ## theta[4] 8.8 6.6 -4.5 4.5 9.2 13.3 20.4 1.0 110 ## theta[5] 6.8 6.9 -8.2 2.3 7.5 11.4 17.7 1.0 410 ## theta[6] 7.3 7.2 -8.6 2.7 8.2 11.8 18.9 1.0 190 ## theta[7] 11.5 6.4 -0.3 7.5 11.2 15.7 25.0 1.1 42 ## theta[8] 9.7 7.6 -4.7 5.1 9.6 14.4 25.1 1.0 130 ## mu.theta 9.2 5.2 -1.2 5.8 9.3 12.5 18.2 1.0 88 ## sigma.theta 5.9 5.6 0.2 1.7 4.4 8.5 20.2 1.1 51 ## deviance 60.7 2.2 57.2 59.2 60.1 61.9 65.6 1.0 120 ## ## For each parameter, n.eff is a crude measure of effective sample size, ## and Rhat is the potential scale reduction factor (at convergence, Rhat=1). ## ## DIC info (using the rule, pD = Dbar-Dhat) ## pD = 2.8 and DIC = 63.4 ## DIC is an estimate of expected predictive error (lower deviance is better). When run natively, WinBUGS and OpenBUGS have nice debugging capabilities; also, you can see what is going on, I mean the program reading the data, generating inits, and so on. To get the OpenBUGS window with a bunch of useful info, just add debug=T to the call of the bugs function, and re-run the model\nschools.sim \u0026lt;- bugs(data, inits, model.file = model.file1,parameters=parameters,n.chains = 3, n.iter = 1000, OpenBUGS.pgm=OpenBUGS.pgm, WINE=WINE, WINEPATH=WINEPATH,useWINE=T,debug=T) ## arguments 'show.output.on.console', 'minimized' and 'invisible' are for Windows only You will have to close the OpenBUGS window to get the prompt back.\n","date":1515801600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1515801600,"objectID":"dd8bf5c2cf56f37350447bdbe0526de5","permalink":"https://oliviergimenez.github.io/blog/run_openbugs_on_mac/","publishdate":"2018-01-13T00:00:00Z","relpermalink":"/blog/run_openbugs_on_mac/","section":"blog","summary":" I had to use the good old `OpenBUGS` for some analyses that cannot be done in `JAGS`. Below are the steps to install `OpenBUGS` then to run it from your Mac either natively or from `R`. This tutorial is an adaptation of [this post](https://sites.google.com/site/mmeclimate/-bayesmet/openbugs-on-mac-os-x) and [that one](http://www.davideagle.org/r-2/bayesian-modeling-using-winbugs-and-openbugs/running-openbugs-on-mac-using-wine).","tags":["bugs","R","rstats","OpenBUGS"],"title":"Run OpenBUGS on a Mac","type":"blog"},{"authors":null,"categories":null,"content":" Our project `Avoid, reduce and compensate mortality risks of the Eurasian Lynx by collision with transport vehicles` aims to build on previous works on lynx population viability, collision risks while crossing terrestrial transport infrastructures and mitigation strategies. Our team will develop an operational tool upon which the technical operators will be able to rely on in the decision-making process for land-use planning. More details [here](https://sites.google.com/view/erclynx/) (in French). This applied research project gathers a group of various actors (public and private; researchers, wildlife managers and carnivore experts; infrastructure managers and road/transport planning experts) with the following objectives:\nMaintaining functional connectivity between favorable lynx habitats and elaborating strategies for avoiding, reducing and compensating collision risks.\nBuilding a tool to reinforce and facilitation the implementation of land-use planning public policies.\nTo achieve these objectives, we will rely on:\nData on collisions (roads, railways) and external structural factors (ITT presence and their characteristics);\nA predictive statistical model to identify the zones with high probability of collision risks;\nA spatially-explicit individual-based population viability model (SEPVA) to predict extinction risks;\nA user-friendly computer program implemented in R. We will follow a companion modeling approach to ensure the smooth adoption and use of this program by the actors.\nThis project will fill a gap between the academic research, the stakes of land planning, the lynx conservation and the needs of ITT managers/operators. We will contribute to the implementation of national public policies (TVB, SRCE/SRADDET, \u0026ldquo;Avoid, Reduce, Compensate\u0026rdquo; doctrine) by testing our approach on pilot sites in the Vosges and Jura areas. This project will benefit from the sharing of experiences and skills from our colleagues in Switzerland and Germany.\n","date":1511827200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1511827200,"objectID":"4b1034113bd5b47e185f15c5e1455b62","permalink":"https://oliviergimenez.github.io/blog/erc_lynx/","publishdate":"2017-11-28T00:00:00Z","relpermalink":"/blog/erc_lynx/","section":"blog","summary":" Our project `Avoid, reduce and compensate mortality risks of the Eurasian Lynx by collision with transport vehicles` aims to build on previous works on lynx population viability, collision risks while crossing terrestrial transport infrastructures and mitigation strategies. Our team will develop an operational tool upon which the technical operators will be able to rely on in the decision-making process for land-use planning. More details [here](https://sites.google.com/view/erclynx/) (in French).","tags":["carnivores","management","lynx","software"],"title":"Terrestrial transport infrastructures, Lynx population viability and mitigation strategies","type":"blog"},{"authors":null,"categories":null,"content":" Here, I illustrate the possibility to use `JAGS` to simulate data with two examples that might be of interest to population ecologists: first a linear regression, second a Cormack-Jolly-Seber capture-recapture model to estimate animal survival (formulated as a state-space model). The code is available from [GitHub](https://github.com/oliviergimenez/simul_with_jags). Recently, I have been struggling with simulating data from complex hierarchical models. After several unsuccessful attempts in R, I remembered the good old times when I was using WinBUGS (more than 10 years already!) and the possibility to simulate data with it. I\u0026rsquo;m using JAGS now, and a quick search in Google with \u0026lsquo;simulating data with jags\u0026rsquo; led me to a complex example and a simple example.\nSimulating data with JAGS is convenient because you can use (almost) the same code for simulation and inference, and you can carry out simulation studies (bias, precision, interval coverage) in the same environment (namely JAGS).\nLinear regression example We first load the packages we need for this tutorial:\nlibrary(R2jags) library(runjags) library(mcmcplots) Then straight to the point, let\u0026rsquo;s generate data from a linear regression model. The trick is to use a data block, have the simplest model block you could think of and pass the parameters as if they were data. Note that it\u0026rsquo;d be possible to use only a model block, see comment here.\ntxtstring \u0026lt;- ' data{ # Likelihood: for (i in 1:N){ y[i] ~ dnorm(mu[i], tau) # tau is precision (1 / variance) mu[i] \u0026lt;- alpha + beta * x[i] } } model{ fake \u0026lt;- 0 } ' Here, alpha and beta are the intercept and slope, tau the precision or the inverse of the variance, y the response variable and x the explanatory variable.\nWe pick some values for the model parameters that we will use as data:\n# parameters for simulations N = 30 # nb of observations x \u0026lt;- 1:N # predictor alpha = 0.5 # intercept beta = 1 # slope sigma \u0026lt;- .1 # residual sd tau \u0026lt;- 1/(sigma*sigma) # precision # parameters are treated as data for the simulation step data\u0026lt;-list(N=N,x=x,alpha=alpha,beta=beta,tau=tau) Now call JAGS; note that we monitor the response variable instead of parameters as we would do when conducting standard inference:\n# run jags out \u0026lt;- run.jags(txtstring, data = data,monitor=c(\u0026quot;y\u0026quot;),sample=1, n.chains=1, summarise=FALSE) ## Compiling rjags model... ## Calling the simulation using the rjags method... ## Note: the model did not require adaptation ## Burning in the model for 4000 iterations... ## Running the model for 1 iterations... ## Simulation complete ## Finished running the simulation The output is a bit messy and needs to be formatted appropriately:\n# reformat the outputs Simulated \u0026lt;- coda::as.mcmc(out) Simulated ## Markov Chain Monte Carlo (MCMC) output: ## Start = 5001 ## End = 5001 ## Thinning interval = 1 ## y[1] y[2] y[3] y[4] y[5] y[6] y[7] y[8] ## 5001 1.288399 2.52408 3.61516 4.583587 5.600675 6.566052 7.593407 8.457497 ## y[9] y[10] y[11] y[12] y[13] y[14] y[15] y[16] ## 5001 9.70847 10.38035 11.5105 12.55048 13.49143 14.46356 15.45641 16.56148 ## y[17] y[18] y[19] y[20] y[21] y[22] y[23] ## 5001 17.50935 18.51501 19.66197 20.49477 21.57079 22.6199 23.48232 ## y[24] y[25] y[26] y[27] y[28] y[29] y[30] ## 5001 24.57923 25.47368 26.33674 27.46525 28.35525 29.60279 30.42952 dim(Simulated) ## [1] 1 30 dat = as.vector(Simulated) dat ## [1] 1.288399 2.524080 3.615160 4.583587 5.600675 6.566052 7.593407 ## [8] 8.457497 9.708470 10.380351 11.510500 12.550482 13.491435 14.463564 ## [15] 15.456410 16.561483 17.509350 18.515005 19.661969 20.494767 21.570790 ## [22] 22.619899 23.482317 24.579228 25.473676 26.336736 27.465251 28.355248 ## [29] 29.602791 30.429517 Now let\u0026rsquo;s fit the model we used to simulate to the data we just generated. I won\u0026rsquo;t go into the details and assume that the reader is familiar with JAGS and linear regression.\n# specify model in BUGS language model \u0026lt;- paste(\u0026quot;\tmodel { # Likelihood: for (i in 1:N){ y[i] ~ dnorm(mu[i], tau) # tau is precision (1 / variance) mu[i] \u0026lt;- alpha + beta * x[i] } # Priors: alpha ~ dnorm(0, 0.01) # intercept beta ~ dnorm(0, 0.01) # slope sigma ~ dunif(0, 100) # standard deviation tau \u0026lt;- 1 / (sigma * sigma) } \u0026quot;) writeLines(model,\u0026quot;lin_reg.jags\u0026quot;)\t# data jags.data \u0026lt;- list(y = dat, N = length(dat), x = x) # initial values inits \u0026lt;- function(){list(alpha = rnorm(1), beta = rnorm(1), sigma = runif(1,0,10))} # parameters monitored parameters \u0026lt;- c(\u0026quot;alpha\u0026quot;, \u0026quot;beta\u0026quot;, \u0026quot;sigma\u0026quot;) # MCMC settings ni \u0026lt;- 10000 nt \u0026lt;- 6 nb \u0026lt;- 5000 nc \u0026lt;- 2 # call JAGS from R res \u0026lt;- jags(jags.data, inits, parameters, \u0026quot;lin_reg.jags\u0026quot;, n.chains = nc, n.thin = nt, n.iter = ni, n.burnin = nb, working.directory = getwd()) ## module glm loaded ## Compiling model graph ## Resolving undeclared variables ## Allocating nodes ## Graph information: ## Observed stochastic nodes: 30 ## Unobserved stochastic nodes: 3 ## Total graph size: 130 ## ## Initializing model Let\u0026rsquo;s have a look to the results and compare with the parameters we used to simulate the data (see above):\n# summarize posteriors print(res, digits = 3) ## Inference for Bugs model at \u0026quot;lin_reg.jags\u0026quot;, fit using jags, ## 2 chains, each with 10000 iterations (first 5000 discarded), n.thin = 6 ## n.sims = 1668 iterations saved ## mu.vect sd.vect 2.5% 25% 50% 75% 97.5% Rhat ## alpha 0.544 0.038 0.469 0.518 0.545 0.570 0.617 1.000 ## beta 0.998 0.002 0.994 0.997 0.998 1.000 1.003 1.001 ## sigma 0.102 0.015 0.078 0.091 0.100 0.110 0.138 1.002 ## deviance -53.810 2.724 -56.867 -55.808 -54.516 -52.641 -46.676 1.001 ## n.eff ## alpha 1700 ## beta 1700 ## sigma 780 ## deviance 1700 ## ## For each parameter, n.eff is a crude measure of effective sample size, ## and Rhat is the potential scale reduction factor (at convergence, Rhat=1). ## ## DIC info (using the rule, pD = var(deviance)/2) ## pD = 3.7 and DIC = -50.1 ## DIC is an estimate of expected predictive error (lower deviance is better). Pretty close!\nCheck convergence:\n# trace plots traplot(res,c(\u0026quot;alpha\u0026quot;, \u0026quot;beta\u0026quot;, \u0026quot;sigma\u0026quot;)) Plot the posterior distribution of the regression parameters and residual standard deviation:\n# posterior distributions denplot(res,c(\u0026quot;alpha\u0026quot;, \u0026quot;beta\u0026quot;, \u0026quot;sigma\u0026quot;)) Capture-recapture example I now illustrate the use of JAGS to simulate data from a Cormack-Jolly-Seber model with constant survival and recapture probabilities. I assume that the reader is familiar with this model and its formulation as a state-space model.\nLet\u0026rsquo;s simulate!\ntxtstring \u0026lt;- ' data{ # Constant survival and recapture probabilities for (i in 1:nind){ for (t in f[i]:(n.occasions-1)){ phi[i,t] \u0026lt;- mean.phi p[i,t] \u0026lt;- mean.p } #t } #i # Likelihood for (i in 1:nind){ # Define latent state and obs at first capture z[i,f[i]] \u0026lt;- 1 mu2[i,1] \u0026lt;- 1 * z[i,f[i]] # detection is 1 at first capture (\u0026quot;conditional on first capture\u0026quot;) y[i,1] ~ dbern(mu2[i,1]) # then deal w/ subsequent occasions for (t in (f[i]+1):n.occasions){ # State process z[i,t] ~ dbern(mu1[i,t]) mu1[i,t] \u0026lt;- phi[i,t-1] * z[i,t-1] # Observation process y[i,t] ~ dbern(mu2[i,t]) mu2[i,t] \u0026lt;- p[i,t-1] * z[i,t] } #t } #i } model{ fake \u0026lt;- 0 } ' Let\u0026rsquo;s pick some values for parameters and store them in a data list:\n# parameter for simulations n.occasions = 10 # nb of occasions nind = 100 # nb of individuals mean.phi \u0026lt;- 0.8 # survival mean.p \u0026lt;- 0.6 # recapture f = rep(1,nind) # date of first capture data\u0026lt;-list(n.occasions = n.occasions, mean.phi = mean.phi, mean.p = mean.p, f = f, nind = nind) Now run JAGS:\nout \u0026lt;- run.jags(txtstring, data = data,monitor=c(\u0026quot;y\u0026quot;),sample=1, n.chains=1, summarise=FALSE) ## Compiling rjags model... ## Calling the simulation using the rjags method... ## Note: the model did not require adaptation ## Burning in the model for 4000 iterations... ## Running the model for 1 iterations... ## Simulation complete ## Finished running the simulation Format the output:\nSimulated \u0026lt;- coda::as.mcmc(out) dim(Simulated) ## [1] 1 1000 dat = matrix(Simulated,nrow=nind) head(dat) ## [,1] [,2] [,3] [,4] [,5] [,6] [,7] [,8] [,9] [,10] ## [1,] 1 1 0 0 0 0 0 0 0 0 ## [2,] 1 1 1 1 0 0 0 0 0 0 ## [3,] 1 0 0 0 0 0 0 0 0 0 ## [4,] 1 0 0 0 0 0 0 0 0 0 ## [5,] 1 0 0 0 0 0 0 0 0 0 ## [6,] 1 1 1 1 0 0 1 0 1 1 Here I monitored only the detections and non-detections, but it is also possible to get the simulated values for the states, i.e. whether an individual is alive or dead at each occasion. You just need to amend the call to JAGS with monitor=c(\u0026quot;y\u0026quot;,\u0026quot;x\u0026quot;) and to amend the output accordingly.\nNow we fit a Cormack-Jolly-Seber model to the data we\u0026rsquo;ve just simulated, assuming constant parameters:\nmodel \u0026lt;- paste(\u0026quot;\tmodel { # Priors and constraints for (i in 1:nind){ for (t in f[i]:(n.occasions-1)){ phi[i,t] \u0026lt;- mean.phi p[i,t] \u0026lt;- mean.p } #t } #i mean.phi ~ dunif(0, 1) # Prior for mean survival mean.p ~ dunif(0, 1) # Prior for mean recapture # Likelihood for (i in 1:nind){ # Define latent state at first capture z[i,f[i]] \u0026lt;- 1 for (t in (f[i]+1):n.occasions){ # State process z[i,t] ~ dbern(mu1[i,t]) mu1[i,t] \u0026lt;- phi[i,t-1] * z[i,t-1] # Observation process y[i,t] ~ dbern(mu2[i,t]) mu2[i,t] \u0026lt;- p[i,t-1] * z[i,t] } #t } #i } \u0026quot;) writeLines(model,\u0026quot;cjs.jags\u0026quot;)\tPrepare the data:\n# vector with occasion of marking get.first \u0026lt;- function(x) min(which(x!=0)) f \u0026lt;- apply(dat, 1, get.first) # data jags.data \u0026lt;- list(y = dat, f = f, nind = dim(dat)[1], n.occasions = dim(dat)[2]) # Initial values known.state.cjs \u0026lt;- function(ch){ state \u0026lt;- ch for (i in 1:dim(ch)[1]){ n1 \u0026lt;- min(which(ch[i,]==1)) n2 \u0026lt;- max(which(ch[i,]==1)) state[i,n1:n2] \u0026lt;- 1 state[i,n1] \u0026lt;- NA } state[state==0] \u0026lt;- NA return(state) } inits \u0026lt;- function(){list(mean.phi = runif(1, 0, 1), mean.p = runif(1, 0, 1), z = known.state.cjs(dat))} We\u0026rsquo;d like to carry out inference about survival and recapture probabilities:\nparameters \u0026lt;- c(\u0026quot;mean.phi\u0026quot;, \u0026quot;mean.p\u0026quot;) Standard MCMC settings:\nni \u0026lt;- 10000 nt \u0026lt;- 6 nb \u0026lt;- 5000 nc \u0026lt;- 2 Ready to run JAGS!\n# Call JAGS from R (BRT 1 min) cjs \u0026lt;- jags(jags.data, inits, parameters, \u0026quot;cjs.jags\u0026quot;, n.chains = nc, n.thin = nt, n.iter = ni, n.burnin = nb, working.directory = getwd()) ## Compiling model graph ## Resolving undeclared variables ## Allocating nodes ## Graph information: ## Observed stochastic nodes: 900 ## Unobserved stochastic nodes: 902 ## Total graph size: 3707 ## ## Initializing model Summarize posteriors and compare to the values we used to simulate the data:\nprint(cjs, digits = 3) ## Inference for Bugs model at \u0026quot;cjs.jags\u0026quot;, fit using jags, ## 2 chains, each with 10000 iterations (first 5000 discarded), n.thin = 6 ## n.sims = 1668 iterations saved ## mu.vect sd.vect 2.5% 25% 50% 75% 97.5% Rhat ## mean.p 0.596 0.033 0.531 0.574 0.597 0.618 0.660 1.000 ## mean.phi 0.784 0.021 0.742 0.770 0.785 0.799 0.824 1.001 ## deviance 440.611 18.374 408.121 427.569 438.662 452.512 479.608 1.001 ## n.eff ## mean.p 1700 ## mean.phi 1700 ## deviance 1700 ## ## For each parameter, n.eff is a crude measure of effective sample size, ## and Rhat is the potential scale reduction factor (at convergence, Rhat=1). ## ## DIC info (using the rule, pD = var(deviance)/2) ## pD = 168.9 and DIC = 609.5 ## DIC is an estimate of expected predictive error (lower deviance is better). Again pretty close!\nTrace plots\ntraplot(cjs,c(\u0026quot;mean.phi\u0026quot;, \u0026quot;mean.p\u0026quot;)) Posterior distribution plots:\ndenplot(cjs,c(\u0026quot;mean.phi\u0026quot;, \u0026quot;mean.p\u0026quot;)) ","date":1511395200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1511395200,"objectID":"d4d9b9c66c44b9abcde82ffc038aa760","permalink":"https://oliviergimenez.github.io/blog/sim_with_jags/","publishdate":"2017-11-23T00:00:00Z","relpermalink":"/blog/sim_with_jags/","section":"blog","summary":" Here, I illustrate the possibility to use `JAGS` to simulate data with two examples that might be of interest to population ecologists: first a linear regression, second a Cormack-Jolly-Seber capture-recapture model to estimate animal survival (formulated as a state-space model). The code is available from [GitHub](https://github.com/oliviergimenez/simul_with_jags).","tags":["rstats","R","JAGS","simulations"],"title":"Simulating data with JAGS","type":"blog"},{"authors":null,"categories":null,"content":" Following my recent attempt to [fit a HMM model to capture-recapture data with TMB](https://oliviergimenez.github.io/post/multievent_in_tmb/) and the rather estonishing outcome (the code was \u003e 300 time faster than the equivalent R code!), I was curious to add TMB to the [list of options I tried to fit dynamic occupancy models](https://oliviergimenez.github.io/post/occupancy_in_admb/). Well, the least I can say is that TMB is fast, damn fast! The reasons for trying TMB were the same as before: TMB is said to be fast, allows for parallel computations, works with R, accomodates spatial stuff, allows easy implementation of random effects).\nI found materials on the internet to teach myself TMB, at least what I needed to implement a simple HMM model. See here for a linear regression and a Gompertz state space model examples, here for the same linear regression example on Youtube (that\u0026rsquo;s awesome!) and many other examples here.\nThe R code is available on my GitHub here. TMB was\u0026hellip; wait for it\u0026hellip; \u0026gt; 300 times faster than ADMB, \u0026gt; 140 times than Unmarked and \u0026gt; 6000 times faster than Jags (although the comparison with the latter is a bit unfair I suppose). The results are available here.\nI\u0026rsquo;m new to TMB, but I\u0026rsquo;m gonna definitely dig into it. Congrats to the developers! Check out the TMB website as well as the paper that comes with it.\n","date":1503576000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1503576000,"objectID":"2fab12daada98d56b5bcac98e364aa1b","permalink":"https://oliviergimenez.github.io/blog/occupancy_in_tmb/","publishdate":"2017-08-24T12:00:00Z","relpermalink":"/blog/occupancy_in_tmb/","section":"blog","summary":" Following my recent attempt to [fit a HMM model to capture-recapture data with TMB](https://oliviergimenez.github.io/post/multievent_in_tmb/) and the rather estonishing outcome (the code was \u003e 300 time faster than the equivalent R code!), I was curious to add TMB to the [list of options I tried to fit dynamic occupancy models](https://oliviergimenez.github.io/post/occupancy_in_admb/). Well, the least I can say is that TMB is fast, damn fast!","tags":["occupancy","R","hmm","TMB"],"title":"Fitting dynamic occupancy models with TMB","type":"blog"},{"authors":null,"categories":null,"content":" Following my attempts to fit a HMM model to [capture-recapture data with Rcpp](http://localhost:1313/post/multievent_in_rcpp/) and to [occupancy data with ADMB](http://localhost:1313/post/occupancy_in_admb/), a few colleagues suggested TMB as a potential alternative for several reasons (fast, allows for parallel computations, works with R, accomodates spatial stuff, easy implementation of random effects, and probably other reasons that I don't know). I found materials on the internet to teach myself TMB, at least what I needed to implement a simple HMM model. See here for a linear regression and a Gompertz state space model examples, here for the same linear regression example on Youtube (that\u0026rsquo;s awesome!) and many other examples here. However, I got stuck and posted my desperate request for help on the TMB forum. Guess what, I got an answer less than a few hours after - thank you Mollie Brooks!\nThe R code is available on my GitHub here. Minimizing the deviance coded with TMB was\u0026hellip; wait for it\u0026hellip; \u0026gt; 300 times faster than using the deviance coded in standard R.\nHope this is useful.\n","date":1503316800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1503316800,"objectID":"5224317b772b568ef4261a60c1e4fd41","permalink":"https://oliviergimenez.github.io/blog/multievent_in_tmb/","publishdate":"2017-08-21T12:00:00Z","relpermalink":"/blog/multievent_in_tmb/","section":"blog","summary":" Following my attempts to fit a HMM model to [capture-recapture data with Rcpp](http://localhost:1313/post/multievent_in_rcpp/) and to [occupancy data with ADMB](http://localhost:1313/post/occupancy_in_admb/), a few colleagues suggested TMB as a potential alternative for several reasons (fast, allows for parallel computations, works with R, accomodates spatial stuff, easy implementation of random effects, and probably other reasons that I don't know).","tags":["capture-recapture","multievent","R","hmm","TMB"],"title":"Fitting HMM/multievent capture-recapture models with TMB","type":"blog"},{"authors":null,"categories":null,"content":" Over the years (age privilege), I sat on several hiring committees for French universities and research institutes. I also have students who are now in the position to apply for such positions. It had me thinking about what works and what doesn't work in job applications and interviews. Here are my 2 cents suggestions. You can find plenty of more clever and relevant recommendations on the internet (see here, here or here for example). Also, be aware that these recommendations are specific to the French system, with little relevance for other countries.\nYour application\nRead and stick to the guidelines (not that obvious sometimes). The French academic system has some specific requirements to apply, in particular for faculty positions. Try to connect your CV with the job profile.\nDo not hesitate to include figures and boxes in the document. Structure does not hurt, well articulated sections are welcome. Be crystal clear on your current position. Synthetic tables summarizing the courses you taught, the talks you gave, the papers you published will help the referees to grab what is important in a minute (they obviously have more than your application to evaluate). Distinguish papers from oral presentations.\nSubmitted papers do not \u0026lsquo;count\u0026rsquo;. Use preprint servers such as arXiv or bioRxiv so that your referees get a chance to read what you\u0026rsquo;ve been working on lately (and judge the quality by themselves).\nProject yourself in the team/lab where the researcher will be hosted, and in the job itself. Feel free to get in touch with the host team/lab. Try to meet people beforehand. Visit websites. Get to know your hopefully-soon-to-be new lab. Show you\u0026rsquo;re interested in working with your hopefully-soon-to-be new colleagues. This might also help you find out about the \u0026lsquo;dismal job that everyone has been avoiding for years\u0026rsquo;.\nFor lecturer positions, even though it is not required, write a short project section including research objectives and do not forget your teaching objectives!\nHave your applications read by one or two colleagues, if possible someone who has experience in being on the kind of hiring committee you will have to face.\nYour interview\nBe on site the day before ( strikes in France happen).\nAppearances do matter, whether you like it or not, therefore pick something neutral and comfortable (my opinion).\nDo not use slides with too much text or complex figures (no tables please!), and point at your slides so that your visual support is part of your speech. Don\u0026rsquo;t look at the screen or the walls, try to look at the members of the jury. Occupy the space, don\u0026rsquo;t always stay at the same spot, but do not move too much (I\u0026rsquo;m useless at that).\nDo not rush when you talk, and give short answers so that every member of the committee gets a chance to ask you a question. Be honest: if you don\u0026rsquo;t know something, or if you don\u0026rsquo;t understand a question (which most likely means that members of the hiring committee do not understand this question either), say it.\nLast, but not least: rehearse, rehearse and rehearse again, there will be a moment where you do no longer sound like you know your text by heart, but rather, like saying it has become natural.\nFrom my experience, the round of questions is more determinant than the talk itself, but we have very rare opportunities to train on how to answer questions. Try to have sessions of questions with your colleagues to get better at it; it comes with experience, no secret.\nFor short talks in general, my advice is to write your text, at least for the first slides so that you\u0026rsquo;re confident you won\u0026rsquo;t get stuck when you start giving your speech. Make a video of yourself rehearsing and watch it; this is painful (I know by experience), but helpful.\nShow that you\u0026rsquo;re excited by the job offer, and that you like teaching and doing research. At the end of the day, I (sadly) vividly remember only a few candidates, most often those who demonstrated interest in the job and pleasure in doing what they do. I wasn\u0026rsquo;t good at inspiring that myself, but one thing that helped me was to think hard why I was applying, and to convince myself that I wanted the job so badly that I had to convince the committee that they should offer the job to me.\nLast but not least, do not do the job of the committee for the committee. In other words, do not censor yourself. Apply, and let the referees do their job.\nGood luck!\n","date":1502712000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1502712000,"objectID":"67174cc607328aa6d1a8b49954e4f365","permalink":"https://oliviergimenez.github.io/blog/interview_inra/","publishdate":"2017-08-14T12:00:00Z","relpermalink":"/blog/interview_inra/","section":"blog","summary":" Over the years (age privilege), I sat on several hiring committees for French universities and research institutes. I also have students who are now in the position to apply for such positions. It had me thinking about what works and what doesn't work in job applications and interviews. Here are my 2 cents suggestions.","tags":["interview","tenure","academia","university","application"],"title":"Applying and interviewing for an academic job","type":"blog"},{"authors":null,"categories":null,"content":"","date":1502496060,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1502496060,"objectID":"66f56322f3bd540dc2c42b4bbb3ff06a","permalink":"https://oliviergimenez.github.io/my-project/external-project/interact/","publishdate":"2017-08-12T00:01:00Z","relpermalink":"/my-project/external-project/interact/","section":"my-project","summary":"Combining ecology and social sciences to study interactions between human activities and bottlenose dolphins in the French Mediterranean sea.","tags":["sociology","cetaceans","management","interdisciplinary"],"title":"Human-dolphin interactions","type":"my-project"},{"authors":null,"categories":null,"content":"","date":1502496000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1502496000,"objectID":"2499b7cc7137f7de5df3181891fd4f9b","permalink":"https://oliviergimenez.github.io/my-project/external-project/democom/","publishdate":"2017-08-12T00:00:00Z","relpermalink":"/my-project/external-project/democom/","section":"my-project","summary":"Developing multispecies demography to assess the effects of climate and management on community dynamics.","tags":["demography","ANR","management","climate"],"title":"Multispecies demography","type":"my-project"},{"authors":null,"categories":null,"content":" Following my previous post on [using ADMB to fit hidden Markov models](https://oliviergimenez.github.io/post/occupancy_in_admb/), I took some time to learn how to use Rcpp ([Eddelbuettel \u0026 Francois 2011](https://www.jstatsoft.org/article/view/v040i08); [Eddelbuettel 2013](http://www.springer.com/us/book/9781461468677)), a package that gives friendly access to the power of C++ and increase the speed of your R programs. Kudos to Dirk Eddelbuettel, Romain Francois and their colleagues, Rcpp is awesome! I started with the excellent Rcpp chapter in the Advanced R book by Hadley Wickham which I complemented with the various vignettes that come with the package. As always, I googled the problems I had and often ended up finding the solution on stackoverflow. The rcpp-devel discussion list is the place where questions should be asked about Rcpp.\nMy objective was to implement the likelihood of a relatively simple multievent capture-recapture model ( Pradel 2005) with Rcpp. I recycled some R code I had and a dataset on shearwaters I used in a paper ( Gimenez et al. 2012).\nThe R code is available on my GitHub here. To run it, you just need to type Rcpp::sourceCpp(\u0026lsquo;multi event.cpp\u0026rsquo;) in the console. I\u0026rsquo;m convinced that the code can be improved, but this simple exercise showed that minimizing the deviance coded with Rcpp and calculating the Hessian was 10 times faster than using the deviance coded in standard R.\nNext steps will be to go for RcppArmadillo for matrix computations and RcppNumerical for optimisation (and numerical integration for random effects).\nHope this is useful.\nUpdate: Following an advice from Romain Francois and Dirk Eddelbuettel (the Rcpp gurus), I have switched to RcppArmadillo to rely on the code developed by professionals and decades of testing. Now the RcppArmadillo code is 50 times faster than basic R! The code is available on my GitHub.\n","date":1502452800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1502452800,"objectID":"f5fa1210c972854031fd4113fe0c2471","permalink":"https://oliviergimenez.github.io/blog/multievent_in_rcpp/","publishdate":"2017-08-11T12:00:00Z","relpermalink":"/blog/multievent_in_rcpp/","section":"blog","summary":" Following my previous post on [using ADMB to fit hidden Markov models](https://oliviergimenez.github.io/post/occupancy_in_admb/), I took some time to learn how to use Rcpp ([Eddelbuettel \u0026 Francois 2011](https://www.jstatsoft.org/article/view/v040i08); [Eddelbuettel 2013](http://www.springer.com/us/book/9781461468677)), a package that gives friendly access to the power of C++ and increase the speed of your R programs. Kudos to Dirk Eddelbuettel, Romain Francois and their colleagues, Rcpp is awesome!","tags":["capture-recapture","multievent","R","hmm"],"title":"Fitting multievent capture-recapture models with Rcpp","type":"blog"},{"authors":null,"categories":null,"content":" Some time ago, a student of mine got stuck when fitting dynamic occupancy models to real data in Jags because of the computational burden. We had a dataset with several thousands sites, more than 20 seasons and 4 surveys per season (yeah!).\nWe thought of using Unmarked instead (the likelihood is written in C++ and used through Rcpp), but dynamic models with false positives and/or random effects are not (yet?) implemented, and we were interested in considering both in our analysis. Some years ago, I had the opportunity to learn ADMB in a NCEAS meeting (thanks Hans Skaug!), I thought I would give it a try. ADMB allows you to write down any likelihood functions yourself and to incorporate random effects in an efficient way. It\u0026rsquo;s known to be fast for reasons I won\u0026rsquo;t go into here. Last but not least, ADMB can be run from R like JAGS and Unmarked (thanks Ben Bolker!).\nHere we go. I first simulate some data, then fit a dynamic model using ADMB, JAGS and Unmarked and finally perform a quick benchmarking. I\u0026rsquo;m going for a standard dynamic model, because the aims are i) to verify that JAGS is slower than Unmarked, ii) that ADMB is closer to Unmarked than JAGS in terms of time computation. If ii) is verified, then it will be worth the effort coding everything in ADMB.\nThe results are available on RPub here. The code is available on GitHub here.\nHope this is useful.\n","date":1502020800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1502020800,"objectID":"ea126ad1d912f19f4c3d55c5b7bb5e1b","permalink":"https://oliviergimenez.github.io/blog/occupancy_in_admb/","publishdate":"2017-08-06T12:00:00Z","relpermalink":"/blog/occupancy_in_admb/","section":"blog","summary":" Some time ago, a student of mine got stuck when fitting dynamic occupancy models to real data in Jags because of the computational burden.","tags":["jags","occupancy","R","hmm"],"title":"Fitting occupancy models in ADMB","type":"blog"},{"authors":null,"categories":null,"content":" I just submitted to Oikos a paper entitled \"Individual heterogeneity and capture-recapture models: what, why and how?\". A pre-print is available [here](http://biorxiv.org/content/early/2017/03/26/120683). I also provide R codes on GitHub [there](http://bit.ly/2nqWzoZ). This is joint work with Emmanuelle Cam and Jean-Michel Gaillard. This work was greatly inspired by my HDR I defended back in 2010 (!) and the ANR project I got over the period 2008-2011 (re-!).\n","date":1490616000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1490616000,"objectID":"bc19b0c3d2bfdb2fd8be08b28f610c23","permalink":"https://oliviergimenez.github.io/blog/paper_idh/","publishdate":"2017-03-27T12:00:00Z","relpermalink":"/blog/paper_idh/","section":"blog","summary":" I just submitted to Oikos a paper entitled \"Individual heterogeneity and capture-recapture models: what, why and how?\". A pre-print is available [here](http://biorxiv.org/content/early/2017/03/26/120683). I also provide R codes on GitHub [there](http://bit.ly/2nqWzoZ).","tags":["paper","individual heterogeneity","capture-recapture"],"title":"Review on individual heterogeneity in capture-recapture models","type":"blog"},{"authors":null,"categories":null,"content":" Our report (in French) on foresighting wolf population dynamic in France is available from [here](https://t.co/a6ibQoF7Nr). My contribution was mostly based on the work of my present or former students, including L. Marescot, J. Louvrier and S. Cubaynes. Difficult exercise, partly because you constantly think of what people will retain from your writings.\n","date":1490443200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1490443200,"objectID":"937ba394171c2cbb64a75b5a0b66217b","permalink":"https://oliviergimenez.github.io/blog/wolf_expertise/","publishdate":"2017-03-25T12:00:00Z","relpermalink":"/blog/wolf_expertise/","section":"blog","summary":" Our report (in French) on foresighting wolf population dynamic in France is available from [here](https://t.co/a6ibQoF7Nr). My contribution was mostly based on the work of my present or former students, including L. Marescot, J. Louvrier and S. Cubaynes.","tags":["expertise","wolf","population dynamics"],"title":"Expertise on wolf population dynamic in France","type":"blog"},{"authors":null,"categories":null,"content":" We held our annual 5-day [capture-recapture workshop](https://multievent.sciencesconf.org/) in Montpellier. As always, the attendees had challenging questions and stimulating datasets to analyse! Looking forward to next year. ","date":1490011200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1490011200,"objectID":"e757f1fb972fe1bc3fba8985e8ebfb01","permalink":"https://oliviergimenez.github.io/blog/cr_workshop_2017/","publishdate":"2017-03-20T12:00:00Z","relpermalink":"/blog/cr_workshop_2017/","section":"blog","summary":" We held our annual 5-day [capture-recapture workshop](https://multievent.sciencesconf.org/) in Montpellier. As always, the attendees had challenging questions and stimulating datasets to analyse! Looking forward to next year. ","tags":["workshop","capture-recapture"],"title":"Capture-recapture workshop in Montpellier","type":"blog"},{"authors":null,"categories":null,"content":" Isabelle Arpin, a sociologist from Grenoble, organised a 2-day seminar on interdisciplinarity. The idea was for the participants to share their experience in practicing interdisciplinarity. Coralie Mounet, Nicolas Lescureux and I gave a talk on our collaboration on the attitude towards brown bears.\n","date":1489406400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1489406400,"objectID":"2bfa767f513559115bf7c18d08a7db8b","permalink":"https://oliviergimenez.github.io/blog/seminar_grenoble/","publishdate":"2017-03-13T12:00:00Z","relpermalink":"/blog/seminar_grenoble/","section":"blog","summary":" Isabelle Arpin, a sociologist from Grenoble, organised a 2-day seminar on interdisciplinarity. The idea was for the participants to share their experience in practicing interdisciplinarity.","tags":["conference","interdisciplinarity","social sciences"],"title":"Seminar on interdisciplinarity in practice","type":"blog"},{"authors":null,"categories":null,"content":" I spent 3 days in Berlin at the [Leibniz Institute for Zoo and Wildlife Research](http://www.leibniz-izw.de/welcome.html) catching up with friends and colleagues (L. Marescot, S. Benhaim, S. Kramer-Schadt, A. Courtiol, H. Hofer). We talked about spotted hyenas and monitoring of large carnivores. I also gave a workshop on inferring occupancy using HMM. I even got in touch with an old acquaintance - Steve Beissinger - who happens to be in Berlin for the year. Steve will come and visit us in Montpellier for a few days in May.\n","date":1488974400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1488974400,"objectID":"ea32ac8f546fe118547c2b9b960dcdea","permalink":"https://oliviergimenez.github.io/blog/visit_berlin_2017/","publishdate":"2017-03-08T12:00:00Z","relpermalink":"/blog/visit_berlin_2017/","section":"blog","summary":" I spent 3 days in Berlin at the [Leibniz Institute for Zoo and Wildlife Research](http://www.leibniz-izw.de/welcome.html) catching up with friends and colleagues (L. Marescot, S. Benhaim, S. Kramer-Schadt, A. Courtiol, H. Hofer).","tags":["workshop","Berlin","trip"],"title":"Visit to the IZW Leibniz Institute, Berlin","type":"blog"},{"authors":null,"categories":null,"content":" We organized a meeting between our lab and [Praxiling](http://www.praxiling.fr/?lang=en) a lab working in linguistic to explore potential collaborations between ecology and linguistic. Marie Chandelier and I gave a talk on our recent results on a content analysis we performed on wolf recovery in France and its coverage by newspapers. More soon on this front. ","date":1488196800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1488196800,"objectID":"9aa419d4c242b85e000d835ca0876fb3","permalink":"https://oliviergimenez.github.io/blog/praxiling/","publishdate":"2017-02-27T12:00:00Z","relpermalink":"/blog/praxiling/","section":"blog","summary":"We organized a meeting between our lab and [Praxiling](http://www.praxiling.fr/?lang=en) a lab working in linguistic to explore potential collaborations between ecology and linguistic. Marie Chandelier and I gave a talk on our recent results on a content analysis we performed on wolf recovery in France and its coverage by newspapers.","tags":["seminar","interdisciplinarity","linguistic"],"title":"Ecology meets linguistic","type":"blog"},{"authors":null,"categories":null,"content":"We had our first meeting for the project \u0026ldquo;Climate and management effects on COMmunity dynamics – Developing multi-species DEMOgraphy\u0026quot;funded by ANR. The website is operational and running, check it out here.\n","date":1487851200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1487851200,"objectID":"9f7bf3e86531d3fb613d80d82be2c36a","permalink":"https://oliviergimenez.github.io/blog/anr_democom_kickoff/","publishdate":"2017-02-23T12:00:00Z","relpermalink":"/blog/anr_democom_kickoff/","section":"blog","summary":"We had our first meeting for the project \u0026ldquo;Climate and management effects on COMmunity dynamics – Developing multi-species DEMOgraphy\u0026quot;funded by ANR. The website is operational and running, check it out here.","tags":["grant","project","multispecies","demography"],"title":"DEMOCOM kick-off meeting","type":"blog"},{"authors":null,"categories":null,"content":"We had our first meeting for the project \u0026ldquo;Scenarios for coexistence between men and dolphins on the Mediterranean French littoral\u0026rdquo; funded by Fondation de France. The website is operational and running, check it out here.\n","date":1485777600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1485777600,"objectID":"6b695cbd1c14e9f05b5227c22e9453ee","permalink":"https://oliviergimenez.github.io/blog/interact_kickoff/","publishdate":"2017-01-30T12:00:00Z","relpermalink":"/blog/interact_kickoff/","section":"blog","summary":"We had our first meeting for the project \u0026ldquo;Scenarios for coexistence between men and dolphins on the Mediterranean French littoral\u0026rdquo; funded by Fondation de France. The website is operational and running, check it out here.","tags":["grant","project","human-wildlife interactions","dolphins"],"title":"INTERACT kick-off meeting","type":"blog"},{"authors":null,"categories":null,"content":"We organised a 2-day workshop on the metrics of longevity used in human and animal demography. The idea is to explore potential bridges between the two fields. Next meeting in April. The group is led by S. Cubaynes, a former PhD student of mine, who now holds a lecturer position at Montpellier University.\n","date":1485432000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1485432000,"objectID":"c7ed4d141086ecde8239c7c06e49721f","permalink":"https://oliviergimenez.github.io/blog/working_group_demo/","publishdate":"2017-01-26T12:00:00Z","relpermalink":"/blog/working_group_demo/","section":"blog","summary":"We organised a 2-day workshop on the metrics of longevity used in human and animal demography. The idea is to explore potential bridges between the two fields. Next meeting in April.","tags":["working group","demography","R"],"title":"Working group on animal/human demography","type":"blog"},{"authors":null,"categories":null,"content":" Ian Renner from the University of Newcastle in Australia, who visited us back in 2015, is with us again for a short 3-week visit.\nWe resumed the work we started one year ago on developing a method to combine sources of information on species distribution. We also started a new project on model selection in occupancy and capture-recapture models using the LASSO.\n","date":1483272000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1483272000,"objectID":"6ab329f532b7cb9145eac4d27430b1fc","permalink":"https://oliviergimenez.github.io/blog/ian_visit2/","publishdate":"2017-01-01T12:00:00Z","relpermalink":"/blog/ian_visit2/","section":"blog","summary":" Ian Renner from the University of Newcastle in Australia, who visited us back in 2015, is with us again for a short 3-week visit.\n","tags":["visitor","species distribution models"],"title":"Another visit of Ian Renner","type":"blog"},{"authors":null,"categories":null,"content":"Blaise defended his PhD I co-supervised with Pierre-Yves Quenette from ONCFS (manuscript here). Blaise studied the attitudes of the public toward brown bear presence and provided sound estimates of abundance and distribution for the species in the Pyrenees.\nWe had stimulating discussions with the committee and F. Sarrazin, G. Chapron, C. Mounet, D. Grémillet and A. Molinari.\n","date":1481544000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1481544000,"objectID":"1a065145e2d1151b8ad7d03b7b0cc65c","permalink":"https://oliviergimenez.github.io/blog/blaise_defense/","publishdate":"2016-12-12T12:00:00Z","relpermalink":"/blog/blaise_defense/","section":"blog","summary":"Blaise defended his PhD I co-supervised with Pierre-Yves Quenette from ONCFS (manuscript here). Blaise studied the attitudes of the public toward brown bear presence and provided sound estimates of abundance and distribution for the species in the Pyrenees.\n","tags":["PhD","brown bear","social sciences"],"title":"Blaise defended his PhD on a socio-ecological approach to Brown bear management in the Pyrenees","type":"blog"},{"authors":null,"categories":null,"content":"Frederic Gosselin, Etienne Rivot and I organised a 2-day workshop on issues in fitting hierarchical models in ecology. I gave a talk on \u0026ldquo;Local minima and multistate capture-recapture models\u0026rdquo;, slides and R code available on GitHub.\nThe idea is to consider several problematic case studies and explore the issues using several computing platforms (R, Jags, Nimble, Stan, Admb). Next meeting in May.\n","date":1480334400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1480334400,"objectID":"73f675f4800f48b8e2e59c67f65baf19","permalink":"https://oliviergimenez.github.io/blog/working_group_hm/","publishdate":"2016-11-28T12:00:00Z","relpermalink":"/blog/working_group_hm/","section":"blog","summary":"Frederic Gosselin, Etienne Rivot and I organised a 2-day workshop on issues in fitting hierarchical models in ecology. I gave a talk on \u0026ldquo;Local minima and multistate capture-recapture models\u0026rdquo;, slides and R code available on GitHub.\n","tags":["working group","hierarchical models","R"],"title":"Workshop on issues in fitting hierarchical models in ecology","type":"blog"},{"authors":null,"categories":null,"content":"During the period 2013-2015, the NGO GIS3M led the GDEGeM project in which tons of data (on abundance and distribution among others) were collected on bottlenose dolphins.\nWe had a 1-day seminar gathering participants to the project, local authorities, lay persons who came to listen to the results of the projects. The talks can be seen here. I contributed by presenting our abundance estimates\nand our new project on the coexistence between dolphins and human activities ","date":1478779200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1478779200,"objectID":"c12dbd0aa2e6e11c1b858f924668bd91","permalink":"https://oliviergimenez.github.io/blog/gdegem_restitution/","publishdate":"2016-11-10T12:00:00Z","relpermalink":"/blog/gdegem_restitution/","section":"blog","summary":"During the period 2013-2015, the NGO GIS3M led the GDEGeM project in which tons of data (on abundance and distribution among others) were collected on bottlenose dolphins.\n","tags":["conference","dolphin","capture-recapture","social sciences"],"title":"GDEGeM workshop on bottlenose dolphin conservation in the North-Western Mediterranean Sea","type":"blog"},{"authors":null,"categories":null,"content":"We attended the conference organized by the French Ecological Society in Marseille. Our talk was \u0026lsquo;Conservation and management of large carnivores in France: A beneficial collaboration between ONCFS practitioners and CNRS researchers\u0026rsquo;.\nIt was a contribution to the Invited Symposium \u0026ldquo;Reciprocal benefits between practitioners, researchers and the biodiversity\u0026rdquo;. This was joint work with M. Guinot-Ghestem, E. Marboutin, Pierre-Yves Quenette, Laetitia Blanc, Sarah Cubaynes, Christophe Duchamp, Julie Louvrier, Lucile Marescot and Blaise Piédallu. The abstract is here.\n","date":1477396800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1477396800,"objectID":"fda79420301b8cb21db5ffbc6e9c8bae","permalink":"https://oliviergimenez.github.io/blog/sfe_2016/","publishdate":"2016-10-25T12:00:00Z","relpermalink":"/blog/sfe_2016/","section":"blog","summary":"We attended the conference organized by the French Ecological Society in Marseille. Our talk was \u0026lsquo;Conservation and management of large carnivores in France: A beneficial collaboration between ONCFS practitioners and CNRS researchers\u0026rsquo;.\n","tags":["conference","talk"],"title":"Talk on the science-management gap in Marseille","type":"blog"},{"authors":null,"categories":null,"content":" Raluca Bancila, who visited us back in 2015, is with us again for a short 2-week visit. We will resume a project we started one year ago on the evaluation of demographic parameters of two related species of spadefoot toads (Pelobates syriacus and P. fuscus) from the Danube Delta Biosphere Reserve, Romania, using capture-recapture models.\n","date":1475323200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1475323200,"objectID":"bb8406dcfe9c4eb8089202e631817aeb","permalink":"https://oliviergimenez.github.io/blog/raluca_visit2/","publishdate":"2016-10-01T12:00:00Z","relpermalink":"/blog/raluca_visit2/","section":"blog","summary":"Raluca Bancila, who visited us back in 2015, is with us again for a short 2-week visit. We will resume a project we started one year ago on the evaluation of demographic parameters of two related species of spadefoot toads (Pelobates syriacus and P.","tags":["visitor","capture-recapture"],"title":"Another visit of Raluca Bancila","type":"blog"},{"authors":null,"categories":null,"content":" Laura Cowen from the Department of Math and Stat at University of Victoria (Canada) will be on a sabbatical with us for 3 months. We will be working on the applications of hidden Markov models to capture-recapture data.\n","date":1472731200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1472731200,"objectID":"553ac55f0a6dde441c7017b62348caef","permalink":"https://oliviergimenez.github.io/blog/laura_cowen/","publishdate":"2016-09-01T12:00:00Z","relpermalink":"/blog/laura_cowen/","section":"blog","summary":"Laura Cowen from the Department of Math and Stat at University of Victoria (Canada) will be on a sabbatical with us for 3 months. We will be working on the applications of hidden Markov models to capture-recapture data.","tags":["visitor","hmm"],"title":"Laura Cowen's sabbatical","type":"blog"},{"authors":null,"categories":null,"content":"This is a free adaptation of two (very) clever analyses made by others:\nThe Star Wars Social Network by Evelina Gabasov in which program F# was mostly used to analyse the Star wars social networks\nAnalyzing networks of characters in \u0026lsquo;Love Actually\u0026rsquo; by David Robinson in which R was used to analyse the links between the characters of the movie Love Actually.\nThe aim here is to try and reproduce Evelina\u0026rsquo;s analysis using R only, using David\u0026rsquo;s contribution plus several tweaks I found here and there on the internet. The R code and data are available on my GitHub page.\nDisclaimer: The original blog posts are awesome and full of relevant details, check them out! My objective here was to teach myself how to manipulate data using trendy R packages and do some network analyses. Some comments below have been copied and pasted from these blogs, the credits entirely go to the authors Evelina and David. Last but not least, my code comes with mistakes probably.\nRead and format data First, read in data. I found the movie script in doc format here, which I converted in txt format for convenience. Then, apply various treatments to have the data ready for analysis. I use the old school way for modifying the original dataframe. Piping would have made the code more readable, but I do not feel confident with this approach yet.\n# load convenient packages library(dplyr) library(stringr) library(tidyr) # read file line by line raw \u0026lt;- readLines(\u0026quot;attack-of-the-clones.txt\u0026quot;) # create data frame lines \u0026lt;- data_frame(raw = raw) # get rid of leading and trailing white spaces # http://stackoverflow.com/questions/2261079/how-to-trim-leading-and-trailing-whitespace-in-r trim \u0026lt;- function (x) gsub(\u0026quot;^\\\\s+|\\\\s+$\u0026quot;, \u0026quot;\u0026quot;, x) lines \u0026lt;- mutate(lines,raw=trim(raw)) # get rid of the empty lines lines2 \u0026lt;- filter(lines, raw != \u0026quot;\u0026quot;) # detect scenes: begin by EXT. or INT. lines3 \u0026lt;- mutate(lines2, is_scene = str_detect(raw, \u0026quot;T.\u0026quot;),scene = cumsum(is_scene)) # drop lines that start with EXT. or INT. lines4 \u0026lt;- filter(lines3,!is_scene) # distinguish characters from what they say lines5 \u0026lt;- separate(lines4, raw, c(\u0026quot;speaker\u0026quot;, \u0026quot;dialogue\u0026quot;), sep = \u0026quot;:\u0026quot;, fill = \u0026quot;left\u0026quot;,extra='drop') # read in aliases (from Evelina's post) aliases \u0026lt;- read.table('aliases.csv',sep=',',header=T,colClasses = \u0026quot;character\u0026quot;) aliases$Alias ## [1] \u0026quot;BEN\u0026quot; \u0026quot;SEE-THREEPIO\u0026quot; \u0026quot;THREEPIO\u0026quot; \u0026quot;ARTOO-DETOO\u0026quot; ## [5] \u0026quot;ARTOO\u0026quot; \u0026quot;PALPATINE\u0026quot; \u0026quot;DARTH SIDIOUS\u0026quot; \u0026quot;BAIL\u0026quot; ## [9] \u0026quot;MACE\u0026quot; \u0026quot;WINDU\u0026quot; \u0026quot;MACE-WINDU\u0026quot; \u0026quot;NUTE\u0026quot; ## [13] \u0026quot;AUNT BERU\u0026quot; \u0026quot;DOOKU\u0026quot; \u0026quot;BOBA\u0026quot; \u0026quot;JANGO\u0026quot; ## [17] \u0026quot;PANAKA\u0026quot; \u0026quot;NUTE\u0026quot; \u0026quot;KI-ADI\u0026quot; \u0026quot;BIBBLE\u0026quot; ## [21] \u0026quot;BIB\u0026quot; \u0026quot;CHEWIE\u0026quot; \u0026quot;VADER\u0026quot; aliases$Name ## [1] \u0026quot;OBI-WAN\u0026quot; \u0026quot;C-3PO\u0026quot; \u0026quot;C-3PO\u0026quot; \u0026quot;R2-D2\u0026quot; ## [5] \u0026quot;R2-D2\u0026quot; \u0026quot;EMPEROR\u0026quot; \u0026quot;EMPEROR\u0026quot; \u0026quot;BAIL ORGANA\u0026quot; ## [9] \u0026quot;MACE WINDU\u0026quot; \u0026quot;MACE WINDU\u0026quot; \u0026quot;MACE WINDU\u0026quot; \u0026quot;NUTE GUNRAY\u0026quot; ## [13] \u0026quot;BERU\u0026quot; \u0026quot;COUNT DOOKU\u0026quot; \u0026quot;BOBA FETT\u0026quot; \u0026quot;JANGO FETT\u0026quot; ## [17] \u0026quot;CAPTAIN PANAKA\u0026quot; \u0026quot;NUTE GUNRAY\u0026quot; \u0026quot;KI-ADI-MUNDI\u0026quot; \u0026quot;SIO BIBBLE\u0026quot; ## [21] \u0026quot;BIB FORTUNA\u0026quot; \u0026quot;CHEWBACCA\u0026quot; \u0026quot;DARTH VADER\u0026quot; # assign unique name to characters # http://stackoverflow.com/questions/28593265/is-there-a-function-like-switch-which-works-inside-of-dplyrmutate multipleReplace \u0026lt;- function(x, what, by) { stopifnot(length(what)==length(by)) ind \u0026lt;- match(x, what) ifelse(is.na(ind),x,by[ind]) } lines6 \u0026lt;- mutate(lines5,speaker=multipleReplace(speaker,what=aliases$Alias,by=aliases$Name)) # read in actual names (from Evelina's post) actual.names \u0026lt;- read.csv('characters.csv',header=F,colClasses = \u0026quot;character\u0026quot;) actual.names \u0026lt;- c(as.matrix(actual.names)) # filter out non-characters lines7 \u0026lt;- filter(lines6,speaker %in% actual.names) # group by scene lines8 \u0026lt;- group_by(lines7, scene, line = cumsum(!is.na(speaker))) lines9 \u0026lt;- summarize(lines8, speaker = speaker[1], dialogue = str_c(dialogue, collapse = \u0026quot; \u0026quot;)) # Count the lines-per-scene-per-character # Turn the result into a binary speaker-by-scene matrix by_speaker_scene \u0026lt;- count(lines9, scene, speaker) by_speaker_scene ## # A tibble: 447 x 3 ## # Groups: scene [321] ## scene speaker n ## \u0026lt;int\u0026gt; \u0026lt;chr\u0026gt; \u0026lt;int\u0026gt; ## 1 26 PADME 1 ## 2 27 PADME 1 ## 3 29 PADME 1 ## 4 48 PADME 1 ## 5 50 PADME 2 ## 6 66 MACE WINDU 1 ## 7 67 MACE WINDU 1 ## 8 69 YODA 1 ## 9 70 MACE WINDU 1 ## 10 74 YODA 1 ## # ... with 437 more rows library(reshape2) speaker_scene_matrix \u0026lt;-acast(by_speaker_scene , speaker ~ scene, fun.aggregate = length) dim(speaker_scene_matrix) ## [1] 19 321 Analyses Hierarchical clustering norm \u0026lt;- speaker_scene_matrix / rowSums(speaker_scene_matrix) h \u0026lt;- hclust(dist(norm, method = \u0026quot;manhattan\u0026quot;)) plot(h) Timeline Use tree to give an ordering that puts similar characters close together\nordering \u0026lt;- h$labels[h$order] ordering ## [1] \u0026quot;MACE WINDU\u0026quot; \u0026quot;YODA\u0026quot; \u0026quot;SHMI\u0026quot; \u0026quot;QUI-GON\u0026quot; \u0026quot;PLO KOON\u0026quot; ## [6] \u0026quot;LAMA SU\u0026quot; \u0026quot;OBI-WAN\u0026quot; \u0026quot;BAIL ORGANA\u0026quot; \u0026quot;JAR JAR\u0026quot; \u0026quot;POGGLE\u0026quot; ## [11] \u0026quot;ANAKIN\u0026quot; \u0026quot;PADME\u0026quot; \u0026quot;CLIEGG\u0026quot; \u0026quot;BERU\u0026quot; \u0026quot;OWEN\u0026quot; ## [16] \u0026quot;SIO BIBBLE\u0026quot; \u0026quot;RUWEE\u0026quot; \u0026quot;JOBAL\u0026quot; \u0026quot;SOLA\u0026quot; This ordering can be used to make other graphs more informative. For instance, we can visualize a timeline of all scenes:\nscenes \u0026lt;- filter(by_speaker_scene, n() \u0026gt; 1) # scenes with \u0026gt; 1 character scenes2 \u0026lt;- ungroup(scenes) scenes3 \u0026lt;- mutate(scenes2, scene = as.numeric(factor(scene)), character = factor(speaker, levels = ordering)) library(ggplot2) ggplot(scenes3, aes(scene, character)) + geom_point() + geom_path(aes(group = scene)) Create a cooccurence matrix (see here) containing how many times two characters share scenes\ncooccur \u0026lt;- speaker_scene_matrix %*% t(speaker_scene_matrix) heatmap(cooccur) Social network analyses Graphical representation of the network Here the nodes represent characters in the movies. The characters are connected by a link if they both speak in the same scene. And the more the characters speak together, the thicker the link between them.\nlibrary(igraph) g \u0026lt;- graph.adjacency(cooccur, weighted = TRUE, mode = \u0026quot;undirected\u0026quot;, diag = FALSE) plot(g, edge.width = E(g)$weight) Compute standard network features, degree and betweeness.\ndegree(g) ## ANAKIN BAIL ORGANA BERU CLIEGG JAR JAR JOBAL ## 12 1 4 4 4 4 ## LAMA SU MACE WINDU OBI-WAN OWEN PADME PLO KOON ## 1 5 6 4 12 0 ## POGGLE QUI-GON RUWEE SHMI SIO BIBBLE SOLA ## 1 1 4 1 0 4 ## YODA ## 4 betweenness(g) ## ANAKIN BAIL ORGANA BERU CLIEGG JAR JAR JOBAL ## 42.600000 0.000000 1.750000 0.500000 22.000000 0.000000 ## LAMA SU MACE WINDU OBI-WAN OWEN PADME PLO KOON ## 0.000000 18.366667 15.000000 5.250000 55.133333 0.000000 ## POGGLE QUI-GON RUWEE SHMI SIO BIBBLE SOLA ## 0.000000 0.000000 0.700000 0.000000 0.000000 5.000000 ## YODA ## 3.366667 To get a nicer representation of the network, see here and the formating from igraph to d3Network. Below is the code you’d need:\nlibrary(d3Network) library(networkD3) sg \u0026lt;- simplify(g) df \u0026lt;- get.edgelist(g, names=TRUE) df \u0026lt;- as.data.frame(df) colnames(df) \u0026lt;- c('source', 'target') df$value \u0026lt;- rep(1, nrow(df)) # get communities fc \u0026lt;- fastgreedy.community(g) com \u0026lt;- membership(fc) node.info \u0026lt;- data.frame(name=names(com), group=as.vector(com)) links \u0026lt;- data.frame(source=match(df$source, node.info$name)-1,target=match(df$target, node.info$name)-1,value=df$value) forceNetwork(Links = links, Nodes = node.info,Source = \u0026quot;source\u0026quot;, Target = \u0026quot;target\u0026quot;,Value = \u0026quot;value\u0026quot;, NodeID = \u0026quot;name\u0026quot;,Group = \u0026quot;group\u0026quot;, opacity = 1, opacityNoHover=1) The nodes represent characters in the movies. The characters are connected by a link if they both speak in the same scene. The colors are for groups obtained by some algorithms.\n","date":1470571200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1470571200,"objectID":"1d35aafc874b8412f4d1332334926076","permalink":"https://oliviergimenez.github.io/blog/starwars_network/","publishdate":"2016-08-07T12:00:00Z","relpermalink":"/blog/starwars_network/","section":"blog","summary":"This is a free adaptation of two (very) clever analyses made by others:\nThe Star Wars Social Network by Evelina Gabasov in which program F# was mostly used to analyse the Star wars social networks","tags":["star wars","social networks","R","rstats"],"title":"Analysing the social Star Wars network in The Attack of the Clones with R","type":"blog"},{"authors":null,"categories":null,"content":"Wow, another good news, our project \u0026ldquo;Climate and management effects on COMmunity dynamics – Developing multi-species DEMOgraphy\u0026rdquo; will be funded by ANR!\nI submitted it in April for the third time. I originally submitted it (just once) to the ERC back in 2011, it was rejected. Over the years, we received constructive comments from referees that helped improving the project, and made it more solid.\nIn DEMOCOM, we will develop an integrated statistical framework for estimation and inference about community dynamics via the development of a multispecies demography, bringing together concepts, predictions and tools from both fields of demography and community ecology. We will apply this new framework to case studies on fishes, mammals and birds. The methods will be made available through R packages.\nThis is a collaborative project with many partners involved, and people to be hired (stay tuned). The project will start early 2017, more soon.\n","date":1469102400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1469102400,"objectID":"dd87e3141b90b50c723014e96cde02cf","permalink":"https://oliviergimenez.github.io/blog/anr_democom/","publishdate":"2016-07-21T12:00:00Z","relpermalink":"/blog/anr_democom/","section":"blog","summary":"Wow, another good news, our project \u0026ldquo;Climate and management effects on COMmunity dynamics – Developing multi-species DEMOgraphy\u0026rdquo; will be funded by ANR!\n","tags":["grant","project","multispecies","demography"],"title":"Research grant on multispecies demography","type":"blog"},{"authors":null,"categories":null,"content":"I decided to teach myself how to do proper reproducible research. Many reasons to that: save time on the mid/long term, make my analyses open and criticizable, share with others, \u0026hellip;\nThere are tons of resources on the web. I use the tutorials cooked by guru Karl Broman. For people like me who learned S-plus, he has some advice to switch to modern tools. Among others, I try and use more and more RStudio in connection with KnitR and R Markdown and share my codes on GitHub. I am more a Fortran guy, but Rcpp makes me wonder about C++. Version control, piping and the tidy universe are not yet entirely familiar to me, I’m making baby steps (or grandpa steps I should say). Let’s see how it goes.\n","date":1468152000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1468152000,"objectID":"f18772904a9680d4411f3d6c63270c3a","permalink":"https://oliviergimenez.github.io/blog/reproducible_research/","publishdate":"2016-07-10T12:00:00Z","relpermalink":"/blog/reproducible_research/","section":"blog","summary":"I decided to teach myself how to do proper reproducible research. Many reasons to that: save time on the mid/long term, make my analyses open and criticizable, share with others, \u0026hellip;\n","tags":["programming","reproducible research","R"],"title":"Self teaching reproducible research","type":"blog"},{"authors":null,"categories":null,"content":"I attended the International Statistical Ecology Conference in Seattle. I don’t go often to conferences, but I do my best not to miss that one.\nThis year, clearly, hidden Markov models (HMM) were on the lips of everyone. Well, it was cool to realize that I wasn\u0026rsquo;t too \u0026ldquo;has been\u0026rdquo; by giving a 1-day workshop on fitting occupancy models with HMM and program E-SURGE developed in our team.\nI also gave a talk on the use of PCA to deal with many correlated covariated in capture-recapture models.\nI also co-authored a talk that Ian Renner (see below) gave on combining sources of information on species distribution.\n","date":1467374400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1467374400,"objectID":"e78d51d764ab0122d73733e22ea564fd","permalink":"https://oliviergimenez.github.io/blog/isec_2016/","publishdate":"2016-07-01T12:00:00Z","relpermalink":"/blog/isec_2016/","section":"blog","summary":"I attended the International Statistical Ecology Conference in Seattle. I don’t go often to conferences, but I do my best not to miss that one.\n","tags":["conference","isec"],"title":"International Statistical Ecology Conference in Seattle","type":"blog"},{"authors":null,"categories":null,"content":"Great news! Our project on \u0026ldquo;Scenarios for coexistence between men and dolphins on the Mediterranean French littoral\u0026rdquo; will be funded by Fondation de France.\nIn this project, we will propose management strategies for human activities and populations of bottlenose dolphins to coexist. Building on the GDEGeM project, we will adopt an ecological approach to infer the spatio-temporal distribution of dolphins and their interactions with human activities, and a sociological approach to explore the attitudes and perceptions of local actors towards dolphins. We will resort to participatory modeling to build, test and use a model to assess management strategies collectively.\nIn passing, I kind of liked the selection procedure: we first submitted a written application, then were asked to come for an interview. I could answer the questions of the committee and have a direct feeling of what they were thinking of our project.\nWe will start this project officially by the end of the year, more soon.\n","date":1464782400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1464782400,"objectID":"dc566a6c89404b085b6b548390853b49","permalink":"https://oliviergimenez.github.io/blog/interact/","publishdate":"2016-06-01T12:00:00Z","relpermalink":"/blog/interact/","section":"blog","summary":"Great news! Our project on \u0026ldquo;Scenarios for coexistence between men and dolphins on the Mediterranean French littoral\u0026rdquo; will be funded by Fondation de France.\n","tags":["grant","human-wildlife interactions","dolphins"],"title":"Research grant on human-dolphin interactions","type":"blog"},{"authors":null,"categories":null,"content":"My post-doc Sarah Cubaynes and I went to visit our colleague Jon Aars in Tromso, Norway. We are working together on the demography of polar bears using an amazing dataset the Norwegian Polar Institute has been collecting for decades. It was also the opportunity to catch up with Nigel (Gilles) Yoccoz who\u0026rsquo;s being very helpful with this project. Jon took us fishing; I was luckier than with fly-fishing ;-)\n","date":1464609600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1464609600,"objectID":"0750e7a68d62c6e8acc925ff36559a0b","permalink":"https://oliviergimenez.github.io/blog/polar_bear/","publishdate":"2016-05-30T12:00:00Z","relpermalink":"/blog/polar_bear/","section":"blog","summary":"My post-doc Sarah Cubaynes and I went to visit our colleague Jon Aars in Tromso, Norway. We are working together on the demography of polar bears using an amazing dataset the Norwegian Polar Institute has been collecting for decades.","tags":["trip","polar bear","Norway","demography"],"title":"Visiting Tromso, polar bear demography","type":"blog"},{"authors":null,"categories":null,"content":"We organised the second conference of the GdR EcoStat. We spent two lovely days in Montpellier talking about ecological statistics with the hundred people who attended. The talks are available here. If you\u0026rsquo;d like to know more about our acitivities in the GdR, check out our website.\n","date":1464264000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1464264000,"objectID":"7aa7d56683dd1e3f5fa9c96a70e8f8fb","permalink":"https://oliviergimenez.github.io/blog/gdr_ecostat_journee2/","publishdate":"2016-05-26T12:00:00Z","relpermalink":"/blog/gdr_ecostat_journee2/","section":"blog","summary":"We organised the second conference of the GdR EcoStat. We spent two lovely days in Montpellier talking about ecological statistics with the hundred people who attended. The talks are available here.","tags":["conference","statistical ecology"],"title":"GDR EcoStat meeting","type":"blog"},{"authors":null,"categories":null,"content":"Among the papers we published lately, I’d like to emphasize two contributions on the conservation of brown bears with some eminent colleagues:\nKaramanlidis, A.A., M. de Gabriel Hernando, L. Krambokoukis, O. Gimenez (2015). Evidence of a large carnivore population recovery: counting bears in Greece. Journal for Nature Conservation. This is a joint venture with Alex and the Greek Arcturos NGO in which we provide the first reliable abundance estimates for brown bears in Greece.\nBischof, R., H. Broseth, O. Gimenez (2015). Wildlife in a politically divided world: insularism inflates estimates of brown bear abundance. Conservation Letters. Richard had this very clever idea to use spatially-explicit capture-recapture models to assess to what extent abundance estimates are affected by double-counting animals that move across borders - see here for a nice summary of our work.\n","date":1464177600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1464177600,"objectID":"771a611ca4cfc9f8432248d57cd4c544","permalink":"https://oliviergimenez.github.io/blog/papers_bears/","publishdate":"2016-05-25T12:00:00Z","relpermalink":"/blog/papers_bears/","section":"blog","summary":"Among the papers we published lately, I’d like to emphasize two contributions on the conservation of brown bears with some eminent colleagues:\nKaramanlidis, A.A., M. de Gabriel Hernando, L. Krambokoukis, O.","tags":["paper","brown bear"],"title":"Focus on two Brown bear papers","type":"blog"},{"authors":null,"categories":null,"content":"","date":1457784000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1457784000,"objectID":"d51521748827ddfe814f7171510ff75f","permalink":"https://oliviergimenez.github.io/blog/fishing2016/","publishdate":"2016-03-12T12:00:00Z","relpermalink":"/blog/fishing2016/","section":"blog","summary":"","tags":["fishing"],"title":"Fishing season is open ;-)","type":"blog"},{"authors":null,"categories":null,"content":"The GIS3M held an international workshop in Marseille on the conservation of Bottlenose dolphin conservation and monitoring in the North-Western Mediterranean Sea. This was the opportunity to discuss the results of the GDEGeM project on the same topic. I mostly contributed by analysing the tons of photo-IDs that were collected during the program to estimate abundance and analyze the social structure. The slides of my talk:\n","date":1448971200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1448971200,"objectID":"fba7bca8edf18b8cd94ef1e95d57fc36","permalink":"https://oliviergimenez.github.io/blog/gis3m/","publishdate":"2015-12-01T12:00:00Z","relpermalink":"/blog/gis3m/","section":"blog","summary":"The GIS3M held an international workshop in Marseille on the conservation of Bottlenose dolphin conservation and monitoring in the North-Western Mediterranean Sea. This was the opportunity to discuss the results of the GDEGeM project on the same topic.","tags":["conference","dolphins","capture-recapture"],"title":"International workshop on Bottlenose dolphin conservation and monitoring","type":"blog"},{"authors":null,"categories":null,"content":"Jean-Dominique Lebreton went on retirement, and we could not let him go without organizing something in his honor. Jean-Do had a tremendous influence in the field of statistical ecology with important contributions in (among others) multivariate analyses, software developments, population dynamics, and capture-recapture methods. On a personal side, I would simply not be here in science without him. Many friends and colleagues came or sent a message for the occasion. These interventions were filmed, see here and enjoy.\n","date":1447416000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1447416000,"objectID":"2d8ee50aa32bdbc468e8cc8cca149c31","permalink":"https://oliviergimenez.github.io/blog/jdl/","publishdate":"2015-11-13T12:00:00Z","relpermalink":"/blog/jdl/","section":"blog","summary":"Jean-Dominique Lebreton went on retirement, and we could not let him go without organizing something in his honor. Jean-Do had a tremendous influence in the field of statistical ecology with important contributions in (among others) multivariate analyses, software developments, population dynamics, and capture-recapture methods.","tags":["conference","statistical ecology"],"title":"Tribute to Jean-Dominique Lebreton","type":"blog"},{"authors":null,"categories":null,"content":"Taking the opportunity of Ian Renner’s visit, we organized thanks to the GdR EcoStat a conference on the recent advances in the modeling of species distribution. We were lucky to have the cream of researchers in the domain with Julien Papaïx, Joseph Chipperfield, Ian Renner, Joern Pagel, Bob O’Hara, Laura Pollock and Damaris Zurell. The talks are available here.\n","date":1446552000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1446552000,"objectID":"f56c5cadeaa24f1f8ddc4607b1c435db","permalink":"https://oliviergimenez.github.io/blog/sdm_day/","publishdate":"2015-11-03T12:00:00Z","relpermalink":"/blog/sdm_day/","section":"blog","summary":"Taking the opportunity of Ian Renner’s visit, we organized thanks to the GdR EcoStat a conference on the recent advances in the modeling of species distribution. We were lucky to have the cream of researchers in the domain with Julien Papaïx, Joseph Chipperfield, Ian Renner, Joern Pagel, Bob O’Hara, Laura Pollock and Damaris Zurell.","tags":["conference","talk"],"title":"Species distribution model day","type":"blog"},{"authors":null,"categories":null,"content":"Ian Renner from the University of Newcastle in Australia spent some time with us while on a sabbatical. We worked together on developing a method to combine sources of information on species distribution. Ian brought his sound expertise on the Lasso for covariate selection and point processes that provide a nice integrated framework. It was fun to program in R in parallel. I wish I could do that more often. We even managed (credit goes to Ian) to present something at the International Statistical Ecology Conference.\nOn a personal side, Ian has become a friend, his kindness and interest in other people are amazing.\n","date":1445860800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1445860800,"objectID":"2748bb81619a0a5be9cc06b685f03f1e","permalink":"https://oliviergimenez.github.io/blog/ian_first_visit/","publishdate":"2015-10-26T12:00:00Z","relpermalink":"/blog/ian_first_visit/","section":"blog","summary":"Ian Renner from the University of Newcastle in Australia spent some time with us while on a sabbatical. We worked together on developing a method to combine sources of information on species distribution.","tags":["visitor"],"title":"Ian Renner's visit","type":"blog"},{"authors":null,"categories":null,"content":"I had the visit of Raluca Bancila a research scientist from Romania. We investigated spatio-temporal environmental variation in and individual heterogeneity on emigration/immigration rates of the troglophile harvestmen species Paranemastoma silli with capture-recapture models. We used two P. silli populations inhabiting two caves, Closani and Lazului Caves, located in the Mehedinti Mountains in southwestern Romania. Yes, we do not only work on big emblematic mammal species ;-) On a personal side, Raluca is always smiling and laughing – ‘la banane’ like we say in French, it was so good to have her in the team. I’m looking forward to hosting her again!\n","date":1443700800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1443700800,"objectID":"061671e16d815f974e6aa2b17ee6d6e4","permalink":"https://oliviergimenez.github.io/blog/raluca_visit1/","publishdate":"2015-10-01T12:00:00Z","relpermalink":"/blog/raluca_visit1/","section":"blog","summary":"I had the visit of Raluca Bancila a research scientist from Romania. We investigated spatio-temporal environmental variation in and individual heterogeneity on emigration/immigration rates of the troglophile harvestmen species Paranemastoma silli with capture-recapture models.","tags":["visitor","individual heterogeneity","capture-recapture"],"title":"Raluca Bancila's visit","type":"blog"},{"authors":null,"categories":null,"content":"We visited our colleagues from Dolphin Biology and Conservation (Giovanni Bearzi, Silvia Bonizzoni, Nina Santostasi ‎and‎‎ Eva Greene). Lots of discussions on future projects about the estimation of striped, bottlenose and common dolphins abundance, their distribution in the gulf of Corinth (Greece) and their interactions with fishermen. Nina will be visiting us in October-March to work on the abundance project. We also went out for a survey and spotted several groups of striped dolphins, and two common dolphins. Thanks to Giovanni and Silvia for their hospitality and kindness. The beginning of a new adventure in Greece!\n","date":1439380800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1439380800,"objectID":"84c4367d1877f224b9c890e02ddc0a7d","permalink":"https://oliviergimenez.github.io/blog/galaxidi/","publishdate":"2015-08-12T12:00:00Z","relpermalink":"/blog/galaxidi/","section":"blog","summary":"We visited our colleagues from Dolphin Biology and Conservation (Giovanni Bearzi, Silvia Bonizzoni, Nina Santostasi ‎and‎‎ Eva Greene). Lots of discussions on future projects about the estimation of striped, bottlenose and common dolphins abundance, their distribution in the gulf of Corinth (Greece) and their interactions with fishermen.","tags":["trip"],"title":"Dolphins, gulf of Corinth, Greece","type":"blog"},{"authors":null,"categories":null,"content":"I attended the International Congress on Biological Conservation that was held in my home town Montpellier. The students of our team were on fire and talked about their exciting work. Gilles presented a bio-economic model to study the population dynamics of Asian captive elephants, a nice example of how to explicitly incorporate interactions between captive and wild populations in species management.\nJulie talked about her use of opportunistic data to map the colonisation of the wolf in France; this is a neat example of how to generate non-detections by carefully assessing the observation efforts to use dynamic occupancy models.\nAnne-Sophie presented an application of Poisson point process modeling to map the presence of brown bear presence in Greece using citizen science data with possible observer bias.\nBlaise presented his results on a sociological survey to assess the attitudes of local residents towards the presence of Brown bears in the Pyrenees.\nLast, Laetitia had a poster on the use of a spatially explicit individual-based model to explore the best strategies (reintroductions and / or corridors) to restore the population of lynx in the Vosges mountains.\nI gave a talk on a review of the methods to analyse citizen science data. This was part of a symposium (nicely summarised by Muki Haklay here) organised by Karine Princé and Wes Hochachka. Stay tuned, a paper is on its way!\nOverall, the program of this conference was dense, the talks and discussions between sessions were stimulating, I’m looking forward to attending the next one!\n","date":1438516800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1438516800,"objectID":"bce30cc635c46d3c8094f60207d331a6","permalink":"https://oliviergimenez.github.io/blog/iccb2014/","publishdate":"2015-08-02T12:00:00Z","relpermalink":"/blog/iccb2014/","section":"blog","summary":"I attended the International Congress on Biological Conservation that was held in my home town Montpellier. The students of our team were on fire and talked about their exciting work. Gilles presented a bio-economic model to study the population dynamics of Asian captive elephants, a nice example of how to explicitly incorporate interactions between captive and wild populations in species management.","tags":["conference","citizen science"],"title":"International Congress on Biological Conservation","type":"blog"},{"authors":null,"categories":null,"content":"Just submitted a big proposal to the Biodiversa call. It was great to write this project with Hervé Fritz who, without a doubt, knows how to write grant applications! If successful, we\u0026rsquo;ll be working on the social and ecological dimensions of connectivity in lynx habitats. This is joint work with French, Swiss, German, Norwegian and Swedish colleagues. More soon hopefully.\nUpdate: On May 2016, I received the news that our project was rejected. Big disappointment. But hey, that\u0026rsquo;s life, and this project is awesome, we\u0026rsquo;ll find other ways to implement it!\n","date":1437393600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1437393600,"objectID":"f05ec51112367326998e453bb2956fd3","permalink":"https://oliviergimenez.github.io/blog/biodiversa/","publishdate":"2015-07-20T12:00:00Z","relpermalink":"/blog/biodiversa/","section":"blog","summary":"Just submitted a big proposal to the Biodiversa call. It was great to write this project with Hervé Fritz who, without a doubt, knows how to write grant applications! If successful, we\u0026rsquo;ll be working on the social and ecological dimensions of connectivity in lynx habitats.","tags":["grant","lynx","socio-ecological approach"],"title":"Biodiversa grant submission","type":"blog"},{"authors":null,"categories":null,"content":"I had the visit of Sarah Benhaim and Lucile Marescot (a former PhD student of mine), both post-doc at the Leibniz Institute in Berlin supervised by Stephanie Kramer-Schadt. It was an exciting week that we spent discussing and analyzing data on the spotted hyenas from the Serengeti National Park with questions at the interface of demography, sociality and epidemiology. This is joint work with Heribert Hofer and Marion East.\n","date":1435752000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1435752000,"objectID":"cc949ea6a0eb71acf6c052dd3a34ae49","permalink":"https://oliviergimenez.github.io/blog/visit_from_izw/","publishdate":"2015-07-01T12:00:00Z","relpermalink":"/blog/visit_from_izw/","section":"blog","summary":"I had the visit of Sarah Benhaim and Lucile Marescot (a former PhD student of mine), both post-doc at the Leibniz Institute in Berlin supervised by Stephanie Kramer-Schadt. It was an exciting week that we spent discussing and analyzing data on the spotted hyenas from the Serengeti National Park with questions at the interface of demography, sociality and epidemiology.","tags":["visitor"],"title":"Visitors from Berlin","type":"blog"},{"authors":null,"categories":null,"content":"I\u0026rsquo;m so delighted to announce that Julie will start a PhD in October with me, jointly working with Eric Marboutin and Christophe Duchamp from ONCFS. There will be also collaborations with John Linnell from NINA. She\u0026rsquo;ll be studying the various ways to assess the distribution of large carnivores and methods to optimise the associated monitoring protocols.\n","date":1435233600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1435233600,"objectID":"fb37f572f0b4a98b8aa2ad6a3ab38784","permalink":"https://oliviergimenez.github.io/blog/grant_julie/","publishdate":"2015-06-25T12:00:00Z","relpermalink":"/blog/grant_julie/","section":"blog","summary":"I\u0026rsquo;m so delighted to announce that Julie will start a PhD in October with me, jointly working with Eric Marboutin and Christophe Duchamp from ONCFS. There will be also collaborations with John Linnell from NINA.","tags":["student","large carnivores"],"title":"New PhD student on large carnivore monitoring","type":"blog"},{"authors":null,"categories":null,"content":"My 2 master students finished their internships. I was so lucky to have them and work with them. Iago worked on the estimation of abundance using bilateral photo-identification capture-recapture data, with an application to bottlenose dolphins. He made a very neat contribution to the field by greatly improving existing algorithms (here and here) to fit capture-recapture models to these non-standard data. Julie worked on inferring the distribution of wolves in France from the beginning of its recolonization in the early 90’s. Her main contribution was to show that by carefully describing observers’ characteristics and properly quantify the prospection effort, citizen science data can be used to map the range dynamics of species with large dispersal distances and home ranges. Iago will spend next year improving his (already largely above the average) skills in informatics, AI, programming, agent-based modeling\u0026hellip; Hopefully, we’ll work again together. Julie is applying for PhD grants, more soon.\n","date":1433160000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1433160000,"objectID":"573248aab9a36227564ce868eb047d2b","permalink":"https://oliviergimenez.github.io/blog/master_2015/","publishdate":"2015-06-01T12:00:00Z","relpermalink":"/blog/master_2015/","section":"blog","summary":"My 2 master students finished their internships. I was so lucky to have them and work with them. Iago worked on the estimation of abundance using bilateral photo-identification capture-recapture data, with an application to bottlenose dolphins.","tags":["student"],"title":"Awesome master students","type":"blog"},{"authors":null,"categories":null,"content":"We had our annual department day; for the occasion, we invited Anne-Caroline Prévot. Anne-Caro is having a fascinating scientific pathway, as she started with a PhD in bird population dynamics and now she’s working on topics requiring skills in environmental psychology. Google her, you’ll see for yourself, her work is awesome. For this special day, we opted for a particular format, with less science than usual, the idea being to show what’s going on behind the scenes. We had a roundtable on the link between research, managers and citizens. We also had several short unorthodox talks where we tried to analyse why some of our papers/projects were successful, but also why we failed at being read or funded. Interestingly, Rob Salguero-Gomez, a colleague of ours, wrote a nice piece on the topic here. Last, we gathered several persons with various professional profiles so that our students could ask them all the questions they wanted. Overall, I think it went amazingly well, and all the attendees found something to take home with them.\n","date":1432209600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1432209600,"objectID":"e799c8f7876ce05f0bef914af399b852","permalink":"https://oliviergimenez.github.io/blog/department_meeting_2015/","publishdate":"2015-05-21T12:00:00Z","relpermalink":"/blog/department_meeting_2015/","section":"blog","summary":"We had our annual department day; for the occasion, we invited Anne-Caroline Prévot. Anne-Caro is having a fascinating scientific pathway, as she started with a PhD in bird population dynamics and now she’s working on topics requiring skills in environmental psychology.","tags":["CEFE","meeting"],"title":"Biodiversity and conservation department meeting","type":"blog"},{"authors":null,"categories":null,"content":"As one of our always excellent department seminar, we had the chance to welcome Heath Smith from University of Washington, Center for Conservation Biology. Heath, together with his dog Chester with whom I became friend during the talk, is working with detection dogs on all sorts of important questions and organisms in conservation biology. Heath then headed to the Pyrénées with Nathalie Espuno to meet with our colleagues from the ONCFS bear (dream)team with whom we\u0026rsquo;re working on Blaise Piédallu\u0026rsquo;s PhD. They\u0026rsquo;re gonna try this technique there, no doubt the results will be of interest for our work on large carnivores!\n","date":1431518400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1431518400,"objectID":"cb9da998721be2e3cf8a1293a9d5f294","permalink":"https://oliviergimenez.github.io/blog/seminar_detection_dogs/","publishdate":"2015-05-13T12:00:00Z","relpermalink":"/blog/seminar_detection_dogs/","section":"blog","summary":"As one of our always excellent department seminar, we had the chance to welcome Heath Smith from University of Washington, Center for Conservation Biology. Heath, together with his dog Chester with whom I became friend during the talk, is working with detection dogs on all sorts of important questions and organisms in conservation biology.","tags":["seminar"],"title":"H. Smith's talk on detection dogs","type":"blog"},{"authors":null,"categories":null,"content":"I acted as the \u0026ldquo;president\u0026rdquo; on the PhD committee of Sophie Monsarrat who, with her supervisor Ana Rodrigues, kindly invited me. It was about the reconstruction of marine mammal’s historical distribution and abundance using historical data. Sophie did an amazing job during her PhD, have a look to her recent paper on the topic. Her presentation was just awesome, I wish I could give such vivid and fascinating talks, irrespective of the topic. Otherwise, her work reminded me of my current work on monk seals with Alex Karamanlidis from the MoM NGO and the talk I\u0026rsquo;m supposed to give at the ICCB2015 on the analysis of citizen science data. More soon.\n","date":1431000000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1431000000,"objectID":"f59e6bdad5bcde0d11fe7343347c640b","permalink":"https://oliviergimenez.github.io/blog/phd_defense/","publishdate":"2015-05-07T12:00:00Z","relpermalink":"/blog/phd_defense/","section":"blog","summary":"I acted as the \u0026ldquo;president\u0026rdquo; on the PhD committee of Sophie Monsarrat who, with her supervisor Ana Rodrigues, kindly invited me. It was about the reconstruction of marine mammal’s historical distribution and abundance using historical data.","tags":["PhD defense"],"title":"PhD defense on historical ecology","type":"blog"},{"authors":null,"categories":null,"content":"Great news! Gilles Maurer will soon start his PhD on the links between wild and captive populations of Asian elephants using an interdisciplinary approach. This is a joint venture with the Beauval zoo and B. Mulot through the CIFRE framework that funds applied PhDs with non-academic partners.\n","date":1429790400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1429790400,"objectID":"ac08c0e8e45171a96f882f40e37494c5","permalink":"https://oliviergimenez.github.io/blog/phd_grant/","publishdate":"2015-04-23T12:00:00Z","relpermalink":"/blog/phd_grant/","section":"blog","summary":"Great news! Gilles Maurer will soon start his PhD on the links between wild and captive populations of Asian elephants using an interdisciplinary approach. This is a joint venture with the Beauval zoo and B.","tags":["PhD grant","socio-ecological approach"],"title":"PhD project on interactions between wild and captive Asian elephant populations","type":"blog"},{"authors":null,"categories":null,"content":"Paméla Lagrange successfully defended her PhD co-supervised with M. Bélisle from the University of Sherbrooke on the \u0026lsquo;Drivers of survival and breeding dispersal using a capture-recapture framework in Tree Swallows - Québec\u0026rsquo;. The members of the committee (M. Festa-Bianchet, G. Gauthier, C. Barbraud, E. Cam) were unanimously amazed by both the ecological and methodological contributions of Paméla to the field. Congrats Pam!\n","date":1428580800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1428580800,"objectID":"0713f224116855a6ef2cbec7c9ec2541","permalink":"https://oliviergimenez.github.io/blog/phd_defense_pam/","publishdate":"2015-04-09T12:00:00Z","relpermalink":"/blog/phd_defense_pam/","section":"blog","summary":"Paméla Lagrange successfully defended her PhD co-supervised with M. Bélisle from the University of Sherbrooke on the \u0026lsquo;Drivers of survival and breeding dispersal using a capture-recapture framework in Tree Swallows - Québec\u0026rsquo;.","tags":["PhD defense"],"title":"Pamela Lagrange's PhD defense","type":"blog"},{"authors":null,"categories":null,"content":"I had the visit of Nina Santostasi from Italy to work for a few days on the estimation of striped dolphin abundance in the gulf of Corinth (in Greece). Nina is working with Giovanni Bearzi and Silvia Bonizzoni from the NGO Dolphin Biology and Conservation. Hopefully our collaboration will provide elements for the conservation of the species.\n","date":1428408000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1428408000,"objectID":"e27aa09c10dab36749db82fd54bb3c17","permalink":"https://oliviergimenez.github.io/blog/nina_visit/","publishdate":"2015-04-07T12:00:00Z","relpermalink":"/blog/nina_visit/","section":"blog","summary":"I had the visit of Nina Santostasi from Italy to work for a few days on the estimation of striped dolphin abundance in the gulf of Corinth (in Greece). Nina is working with Giovanni Bearzi and Silvia Bonizzoni from the NGO Dolphin Biology and Conservation.","tags":["visitor","abundance"],"title":"Nina Santostasi's visit","type":"blog"},{"authors":null,"categories":null,"content":"As part of the GdR EcoStat and our evolutionary demography group, we had a stimulating workshop on the quantification of individual heterogeneity in survival of wild populations using an approach recently developed by Hal Caswell. Hal attended the workshop with Lotte de Vries her new PhD student who will be leading the study.\n","date":1427976000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1427976000,"objectID":"e9182b7d2be29d2060a549b86b0f450e","permalink":"https://oliviergimenez.github.io/blog/gdr_demography/","publishdate":"2015-04-02T12:00:00Z","relpermalink":"/blog/gdr_demography/","section":"blog","summary":"As part of the GdR EcoStat and our evolutionary demography group, we had a stimulating workshop on the quantification of individual heterogeneity in survival of wild populations using an approach recently developed by Hal Caswell.","tags":["demography","working group","individual heterogeneity"],"title":"Working group in demography","type":"blog"},{"authors":null,"categories":null,"content":"We organised the first conference of the GdR EcoStat. We spent two lovely days in Lyon talking about ecological statistics with the hundred people who attended. The talks are available here. If you\u0026rsquo;d like to know more about our activities in the GdR, check out our website.\n","date":1426161600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1426161600,"objectID":"d3fba201c1b9411a52c8138f325ea53d","permalink":"https://oliviergimenez.github.io/blog/gdr_ecostat_conf/","publishdate":"2015-03-12T12:00:00Z","relpermalink":"/blog/gdr_ecostat_conf/","section":"blog","summary":"We organised the first conference of the GdR EcoStat. We spent two lovely days in Lyon talking about ecological statistics with the hundred people who attended. The talks are available here.","tags":["conference","statistical ecology"],"title":"GDR EcoStat first conference","type":"blog"},{"authors":null,"categories":null,"content":"Laetitia Blanc succesfully defended her PhD on the conservation of lynx in France. Congrats Laeti! She had very nice feedbacks by the members of committee (X. Lambin, E. Marboutin, J. Linnell, M. Schaub and F. Bonadonna). Laetitia will be working for a few months with us on the estimation of lynx abundance in Norway (with J. Linnell) and with ONCFS to finish up a chapter of her thesis.\n","date":1423224000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1423224000,"objectID":"753209f0566e9e933dddd35e8ea5ea8a","permalink":"https://oliviergimenez.github.io/blog/laeti_phd_defense/","publishdate":"2015-02-06T12:00:00Z","relpermalink":"/blog/laeti_phd_defense/","section":"blog","summary":"Laetitia Blanc succesfully defended her PhD on the conservation of lynx in France. Congrats Laeti! She had very nice feedbacks by the members of committee (X. Lambin, E. Marboutin, J. Linnell, M.","tags":["PhD defense","lynx"],"title":"Laetitia Blanc defended her PhD","type":"blog"},{"authors":null,"categories":null,"content":"Welcome to our new master students Iago Bonnici and Julie Louvrier and Tamar Lok our new post-doc.\n","date":1422792000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1422792000,"objectID":"e0430b0ae8f4660274b668f82ad0f4ba","permalink":"https://oliviergimenez.github.io/blog/arrivals/","publishdate":"2015-02-01T12:00:00Z","relpermalink":"/blog/arrivals/","section":"blog","summary":"Welcome to our new master students Iago Bonnici and Julie Louvrier and Tamar Lok our new post-doc.","tags":["students"],"title":"Welcome","type":"blog"},{"authors":null,"categories":null,"content":"Our team has recently joined the department Biodiversity \u0026amp; Conservation at CEFE, a department that I have the pleasure to head now. We have moved offices and are now on the second floor. I have an office with a brand new parquet floor and red walls (that I chose and installed/painted myself, can you believe it?!). We had our office-warming party early March. Do not hesitate to come by and visit our new offices!\n","date":1420891200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1420891200,"objectID":"2f7525a72559153ccd559a45270440d4","permalink":"https://oliviergimenez.github.io/blog/move/","publishdate":"2015-01-10T12:00:00Z","relpermalink":"/blog/move/","section":"blog","summary":"Our team has recently joined the department Biodiversity \u0026amp; Conservation at CEFE, a department that I have the pleasure to head now. We have moved offices and are now on the second floor.","tags":["CEFE"],"title":"Moving upstairs","type":"blog"},{"authors":null,"categories":null,"content":"Our meeting report on ISEC2014 has appeared in Biology Letters just before Christmas eve, nice gift! Check it out, it\u0026rsquo;s in open access.\n","date":1419422400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1419422400,"objectID":"d5b00ddb1349017ce3624ccaf25fc87f","permalink":"https://oliviergimenez.github.io/blog/isec_paper/","publishdate":"2014-12-24T12:00:00Z","relpermalink":"/blog/isec_paper/","section":"blog","summary":"Our meeting report on ISEC2014 has appeared in Biology Letters just before Christmas eve, nice gift! Check it out, it\u0026rsquo;s in open access.","tags":["paper","statistical ecology"],"title":"Statistical ecology comes of age","type":"blog"},{"authors":null,"categories":null,"content":"Jim Sedinger, professor at the University of Nevada (Reno), said goodbye to the team after 3 months with us on a sabbatical. It’s been great to have Jim here analyzing data collected as part of the amazing Black Brant monitoring initiated in the 40s. More soon on our joint project on senescence and individual heterogeneity.\n","date":1418731200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1418731200,"objectID":"323243d8d3ad1d196291c594e2b821cd","permalink":"https://oliviergimenez.github.io/blog/jim_sedinger/","publishdate":"2014-12-16T12:00:00Z","relpermalink":"/blog/jim_sedinger/","section":"blog","summary":"Jim Sedinger, professor at the University of Nevada (Reno), said goodbye to the team after 3 months with us on a sabbatical. It’s been great to have Jim here analyzing data collected as part of the amazing Black Brant monitoring initiated in the 40s.","tags":["visitor"],"title":"Jim Sedinger's sabbatical","type":"blog"},{"authors":null,"categories":null,"content":"I acted as president of the examining committee of Sarah Calba who defended her PhD in epistemology of sciences with the predictions in community ecology as a case study. The thesis was supervised by two clever colleagues of mine, Virginie Maris and Vincent Devictor. Sarah did more than well, and the discussions were lively and the debate vivid. As a general comment, I wish we, as ecologists, had more time as part of our training to think of why and how we conduct our scientific activities. Sarah has a nice paper showing that the identification of patterns and processes linking species diversity to functional or phylogenetic diversity depends on the methodological choices we make as analysts. This reminds me of the discussion about \u0026ldquo;Researcher Degrees of Freedom\u0026rdquo; on Gelman’s blog.\n","date":1418731200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1418731200,"objectID":"d90e67a1fa4a648463a1f893b8cfca70","permalink":"https://oliviergimenez.github.io/blog/sarah_calba/","publishdate":"2014-12-16T12:00:00Z","relpermalink":"/blog/sarah_calba/","section":"blog","summary":"I acted as president of the examining committee of Sarah Calba who defended her PhD in epistemology of sciences with the predictions in community ecology as a case study. The thesis was supervised by two clever colleagues of mine, Virginie Maris and Vincent Devictor.","tags":["PhD defense","social sciences"],"title":"Sarah Calba's PhD defense","type":"blog"},{"authors":null,"categories":null,"content":"We held AppliBugs in Montpellier, with the idea to stimulate exchanges and to share information on Bayesian methods and applications ( program here). The day went well despite a rainstorm falling on Montpellier. All talks were great, and I was particularly interested in Adrien Todeschini\u0026rsquo;s presentation of Biips, a software allowing the implementation of state-space models using particle filtering and a R/BUGS-like syntax.\n","date":1417176000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1417176000,"objectID":"eb4bbb495775e5fe7031508928da4a6e","permalink":"https://oliviergimenez.github.io/blog/applibugs_montpellier/","publishdate":"2014-11-28T12:00:00Z","relpermalink":"/blog/applibugs_montpellier/","section":"blog","summary":"We held AppliBugs in Montpellier, with the idea to stimulate exchanges and to share information on Bayesian methods and applications ( program here). The day went well despite a rainstorm falling on Montpellier.","tags":["applibugs","bayes","conference"],"title":"AppliBugs and applications of Bayesian theory","type":"blog"},{"authors":null,"categories":null,"content":"We had a lovely 2-day workshop in Lyon on individual heterogeneity in demography as part of our research group on Statistical Ecology. I gave a short talk providing an overview of heterogeneous capture-recapture models. We agreed on meeting again early April 2015 to carry out a comparative analysis of individual heterogeneity using the datasets shared by the group and the method developed by Hal Caswell (check it out here).\n","date":1416484800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1416484800,"objectID":"f5e07dec086d05e914d29ff6691be905","permalink":"https://oliviergimenez.github.io/blog/idh_meeting/","publishdate":"2014-11-20T12:00:00Z","relpermalink":"/blog/idh_meeting/","section":"blog","summary":"We had a lovely 2-day workshop in Lyon on individual heterogeneity in demography as part of our research group on Statistical Ecology. I gave a short talk providing an overview of heterogeneous capture-recapture models.","tags":["workshop","individual heterogeneity"],"title":"Workshop on individual heterogeneity","type":"blog"},{"authors":null,"categories":null,"content":"After a month off, I’m back to work starting with a 2-day workshop on citizen sciences I’ve co-organised with Romain Julliard and Pascal Monestiez ( program) as part of our group on Statistical Ecology and on citizen sciences CiSStats. The discussions were stimulating with a nice balance between talks given by data providers on the first day and modelers on the second day. I gave a talk based on Anne-Sophie’s work on mapping brown bear distribution in Greece using opportunistic data. Slides below:\nCheck out below Anne-Sophie\u0026rsquo;s talk she gave at the 2014 International Conference on Bear Research and Management in Thessaloniki (Greece) and at the joint BES-SFE Ecology conference in Lille.\n","date":1413460800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1413460800,"objectID":"1e6664c7af342984f035a720b77e6d19","permalink":"https://oliviergimenez.github.io/blog/cs_meeting/","publishdate":"2014-10-16T12:00:00Z","relpermalink":"/blog/cs_meeting/","section":"blog","summary":"After a month off, I’m back to work starting with a 2-day workshop on citizen sciences I’ve co-organised with Romain Julliard and Pascal Monestiez ( program) as part of our group on Statistical Ecology and on citizen sciences CiSStats.","tags":["workshop","citizen science"],"title":"Workshop on citizen science and statistics","type":"blog"},{"authors":null,"categories":null,"content":"I attended Daniel Turek\u0026rsquo;s talk at ISEC about NIMBLE a neat alternative to WinBUGS and JAGS. It is developed by Perry de Valpine\u0026rsquo;s group at Berkeley and \u0026lsquo;lets you use BUGS models natively in R, program functions that use them, and compile everything via C++ for faster computing\u0026rsquo;. I played around with NIMBLE a bit, here is an example of fitting a classic capture-recapture model to simulated data - thanks to Perry and Daniel for their help! NIMBLE seems a lot faster than its competitors, and much more flexible. I\u0026rsquo;ll continue my investigations with more complex models - stay tuned.\n","date":1404993600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1404993600,"objectID":"0869115bd7d03194d6a945b2759b7623","permalink":"https://oliviergimenez.github.io/blog/nimble/","publishdate":"2014-07-10T12:00:00Z","relpermalink":"/blog/nimble/","section":"blog","summary":"I attended Daniel Turek\u0026rsquo;s talk at ISEC about NIMBLE a neat alternative to WinBUGS and JAGS. It is developed by Perry de Valpine\u0026rsquo;s group at Berkeley and \u0026lsquo;lets you use BUGS models natively in R, program functions that use them, and compile everything via C++ for faster computing\u0026rsquo;.","tags":["isec","R","conference"],"title":"New kids on the Bayesian block","type":"blog"},{"authors":null,"categories":null,"content":"ISEC 2014 is now closed. Organising this conference has been a fantastic adventure; hopefully the participants liked it.\n","date":1404561600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1404561600,"objectID":"30ad5fd73edef01a3cb2561964b0f1f1","permalink":"https://oliviergimenez.github.io/blog/isec2014_end/","publishdate":"2014-07-05T12:00:00Z","relpermalink":"/blog/isec2014_end/","section":"blog","summary":"ISEC 2014 is now closed. Organising this conference has been a fantastic adventure; hopefully the participants liked it.","tags":["isec","conference"],"title":"ISEC 2014: the end","type":"blog"},{"authors":null,"categories":null,"content":"I gave a talk at ISEC 2014 on fitting occupancy models with hidden-Markov models using E-SURGE:\n","date":1404561600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1404561600,"objectID":"fd3bf509d9243080e128d52f4f50c0ad","permalink":"https://oliviergimenez.github.io/blog/isec2014_talks/","publishdate":"2014-07-05T12:00:00Z","relpermalink":"/blog/isec2014_talks/","section":"blog","summary":"I gave a talk at ISEC 2014 on fitting occupancy models with hidden-Markov models using E-SURGE:","tags":["isec","conference"],"title":"We're presenting our work at ISEC 2014","type":"blog"},{"authors":null,"categories":null,"content":"Ben Rashford and Thomas Foulke from the University of Wyoming visited us. They gave a great talk on \u0026ldquo;Wolf Re-Introduction in the Northern Rockies, USA: From Population and Predation Trends to Policy and Economics\u0026rdquo;.\n","date":1403697600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1403697600,"objectID":"9f0e47bb3be21f80639f93c9acd2870e","permalink":"https://oliviergimenez.github.io/blog/economy_wolf/","publishdate":"2014-06-25T12:00:00Z","relpermalink":"/blog/economy_wolf/","section":"blog","summary":"Ben Rashford and Thomas Foulke from the University of Wyoming visited us. They gave a great talk on \u0026ldquo;Wolf Re-Introduction in the Northern Rockies, USA: From Population and Predation Trends to Policy and Economics\u0026rdquo;.","tags":["visitor","talk"],"title":"Economy and large carnivores","type":"blog"},{"authors":null,"categories":null,"content":"Rachel Guénon successfully defended her Master internship on the Dalmatian pelican and life-history tradeoffs (from Greece!). This is joint work with Alain Crivelli from Tour du Valat.\n","date":1403352000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1403352000,"objectID":"dc95c25df80bf6f31192a1e86589dbcf","permalink":"https://oliviergimenez.github.io/blog/pelican_rachel/","publishdate":"2014-06-21T12:00:00Z","relpermalink":"/blog/pelican_rachel/","section":"blog","summary":"Rachel Guénon successfully defended her Master internship on the Dalmatian pelican and life-history tradeoffs (from Greece!). This is joint work with Alain Crivelli from Tour du Valat.","tags":["master","defense"],"title":"Pelican and life-history trade-offs","type":"blog"},{"authors":null,"categories":null,"content":"Sending questionnaires for Blaise\u0026rsquo;s PhD to study the local perception of the presence of brown bears in the Pyrénées. First picture, the 3000 questionnaires to be sent; second picture, those questionnaires once packed in boxes; third picture, the nice guy from the mail company La Poste comes and picks them up. Now the question is how many of them we will we get back?\u0026hellip;\n","date":1403265600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1403265600,"objectID":"656a87be874c55a606cb08cd9065131b","permalink":"https://oliviergimenez.github.io/blog/blaise_quest/","publishdate":"2014-06-20T12:00:00Z","relpermalink":"/blog/blaise_quest/","section":"blog","summary":"Sending questionnaires for Blaise\u0026rsquo;s PhD to study the local perception of the presence of brown bears in the Pyrénées. First picture, the 3000 questionnaires to be sent; second picture, those questionnaires once packed in boxes; third picture, the nice guy from the mail company La Poste comes and picks them up.","tags":["questionnaire","social sciences","brown bear","attitude"],"title":"Studying attitudes towards large carnivores","type":"blog"},{"authors":null,"categories":null,"content":"The team is leading the organisation of ISEC2014. We\u0026rsquo;re getting ready by, among many other things, preparing the delegates\u0026rsquo; packs. Of course, these efforts cannot go without some relaxing time.\n","date":1401624000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1401624000,"objectID":"961fbe75d08146fe987ac1666ffdd021","permalink":"https://oliviergimenez.github.io/blog/isec2014_prep/","publishdate":"2014-06-01T12:00:00Z","relpermalink":"/blog/isec2014_prep/","section":"blog","summary":"The team is leading the organisation of ISEC2014. We\u0026rsquo;re getting ready by, among many other things, preparing the delegates\u0026rsquo; packs. Of course, these efforts cannot go without some relaxing time.","tags":["isec","conference"],"title":"We're organising ISEC 2014","type":"blog"},{"authors":null,"categories":null,"content":"Relaxing after finishing up writing a big grant proposal. Tedeschi and Truck concert at the mytic Olympia Paris with Jonny Lang as the opening act.\n","date":1398686400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1398686400,"objectID":"b606bacb0d3a19c4f4962fc166e0afaf","permalink":"https://oliviergimenez.github.io/blog/relax_concert/","publishdate":"2014-04-28T12:00:00Z","relpermalink":"/blog/relax_concert/","section":"blog","summary":"Relaxing after finishing up writing a big grant proposal. Tedeschi and Truck concert at the mytic Olympia Paris with Jonny Lang as the opening act.","tags":["relaxing","music"],"title":"Post grant submission relaxing","type":"blog"},{"authors":null,"categories":null,"content":"Dave and Lise are going back home to the US after a 6-month sabbatical with us. It\u0026rsquo;s been awesome to have you here!\n","date":1397131200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1397131200,"objectID":"95fadecc646087f2dc46aefb0e7fc3f0","permalink":"https://oliviergimenez.github.io/blog/lise_dave/","publishdate":"2014-04-10T12:00:00Z","relpermalink":"/blog/lise_dave/","section":"blog","summary":"Dave and Lise are going back home to the US after a 6-month sabbatical with us. It\u0026rsquo;s been awesome to have you here!","tags":["visitor"],"title":"Bye bye Lise and Dave","type":"blog"},{"authors":null,"categories":null,"content":"The team run two 1-week workshops on capture-recapture models and matrix models. We\u0026rsquo;re grateful to all the attendees who came, with fascinating research topics and stimulating questions. No need to say that your presence means a lot to us! We hope to see again many people next year! Warm and special thanks to Dave Koons, Lise Aubry and Pamela Lagrange!\n","date":1395316800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1395316800,"objectID":"aaf78930310de111ec9f0518f9a2fe8b","permalink":"https://oliviergimenez.github.io/blog/workshop_cmr_2014/","publishdate":"2014-03-20T12:00:00Z","relpermalink":"/blog/workshop_cmr_2014/","section":"blog","summary":"The team run two 1-week workshops on capture-recapture models and matrix models. We\u0026rsquo;re grateful to all the attendees who came, with fascinating research topics and stimulating questions. No need to say that your presence means a lot to us!","tags":["workshop","capture-recapture"],"title":"Capture-recapture workshop","type":"blog"},{"authors":null,"categories":null,"content":"The French research group in Statistical Ecology (GDR EcoStat) is born. Funded by CNRS, the objective is to structure the research in statistical ecology at a national (French) level. Check out our website, and sign up for our mailing list!\n","date":1388577600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1388577600,"objectID":"55b97448b70522e9e90cdd3c31736451","permalink":"https://oliviergimenez.github.io/blog/gdr_ecostat_is_born/","publishdate":"2014-01-01T12:00:00Z","relpermalink":"/blog/gdr_ecostat_is_born/","section":"blog","summary":"The French research group in Statistical Ecology (GDR EcoStat) is born. Funded by CNRS, the objective is to structure the research in statistical ecology at a national (French) level. Check out our website, and sign up for our mailing list!","tags":["statistical ecology"],"title":"The GDR is statistical ecology is born!","type":"blog"},{"authors":null,"categories":null,"content":"Below are the slides of my plenary talk at the 2013 EURING meeting on individual variability in capture-recapture models. I couldn’t attend, but the organizers were kind enough to run my presentation to the audience. I had to be a bit imaginative for the attendees not to fall asleep\u0026hellip;\n","date":1367755200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1367755200,"objectID":"3b97d937868e18c046753f2866b2d155","permalink":"https://oliviergimenez.github.io/blog/euring_2013/","publishdate":"2013-05-05T12:00:00Z","relpermalink":"/blog/euring_2013/","section":"blog","summary":"Below are the slides of my plenary talk at the 2013 EURING meeting on individual variability in capture-recapture models. I couldn’t attend, but the organizers were kind enough to run my presentation to the audience.","tags":["statistical ecology","individual heterogeneity","conference"],"title":"Individual heterogeneity and capture-recapture at EURING 2013","type":"blog"},{"authors":null,"categories":null,"content":"Here are the slides of my interview for a CNRS senior scientist position (DR2). Note that CNRS offers interdisciplinary positions, which is sadly not true for all of our research institutes.\n","date":1334923200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1334923200,"objectID":"91577d3b3c1cd18a77502b695e36f28c","permalink":"https://oliviergimenez.github.io/blog/cnrs_interview/","publishdate":"2012-04-20T12:00:00Z","relpermalink":"/blog/cnrs_interview/","section":"blog","summary":"Here are the slides of my interview for a CNRS senior scientist position (DR2). Note that CNRS offers interdisciplinary positions, which is sadly not true for all of our research institutes.","tags":["cnrs","interview"],"title":"My CNRS interview for a senior scientist position","type":"blog"},{"authors":null,"categories":null,"content":"Here are the slides of my HDR defense. This is a necessary step in France (and in other countries) to be authorized to supervise (officially) PhD students.\n","date":1269777600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1269777600,"objectID":"d47f53e69ec744817d70a56d7fa14d6f","permalink":"https://oliviergimenez.github.io/blog/hdr_defense/","publishdate":"2010-03-28T12:00:00Z","relpermalink":"/blog/hdr_defense/","section":"blog","summary":"Here are the slides of my HDR defense. This is a necessary step in France (and in other countries) to be authorized to supervise (officially) PhD students.","tags":["hdr","defense"],"title":"My HDR defense","type":"blog"},{"authors":null,"categories":null,"content":"Below I provide pieces of code I have written over the years and shared through GitHub on reproducible science, capture-recapture, occupancy and epidemiological models, spatial stuff, social networks, deep learning, textual analyses, the tidyverse and some random stuff.\nThis code is provided as is without warranty of any kind. If you find a bug or anything, feel free to get in touch or fill an issue on GitHub.\nYou may find other pieces of code in papers.\nLast update: August 2021.\nReproducible science Introduction to Git and GitHub with RStudio [ slides] [ material]\nWriting dynamic and reproducible documents with R Markdown [ slides] [ material] [ practical] [ interactive dashboard] [ website]\nManipulate and visualise data in the tidyverse [ slides] [ material] [ tips]\nFocus on GIS and spatial data with sf package [ slides] [ material]\nCapture-recapture Quantifying bias in mortality estimates when ignoring imperfect detection \u0026ndash; R, Shiny\nCapture-recapture models as HMMs \u0026ndash; R, Jags\nSpatial capture-recapture models for closed and open populations \u0026ndash; R, Nimble\nJolly-Seber capture-recapture models \u0026ndash; R, Jags\nMultievent capture-recapture models \u0026ndash; R, Jags, Nimble\nMultievent capture-recapture models with Rcpp \u0026ndash; R, Rcpp\nMultievent SIR models and capture-recapture data \u0026ndash; R, TMB\nLocal minima and multistate capture-recapture models; see also there for mixture of recaptures and recoveries \u0026ndash; R, Jags, ADMB, Stan, unmarked\nAnimal social networks and capture-recapture data \u0026ndash; R, Jags\nIndividual fitness in the wild and capture-recapture data \u0026ndash; R, Jags, E-SURGE\nGompertz capture-recapture model with a Gamma frailty \u0026ndash; R, Jags\nMany correlated covariates in capture-recapture models \u0026ndash; R, RMark\nStructural equation capture-recapture models \u0026ndash; R, Jags\nAbundance estimation using capture-recapture models \u0026ndash; R, RMark\nBand-recovery models with time random effects in a frequentist framework \u0026ndash; R\nBayesian implementation of capture-recapture models with robust design \u0026ndash; R, Jags\nPredator-prey integrated community models \u0026ndash; R, Jags\nOccupancy Dynamic analysis of brown bear habitat use in the Pyrenees \u0026ndash; R, unmarked\nDynamic occupancy models in TMB and ADMB \u0026ndash; R, TMB, ADMB\nMultistate occupancy model with uncertainty \u0026ndash; R, Jags\nHMM formulation of a multispecies dynamic occupancy models \u0026ndash; R\nSimulation and fit of a 2-species occupancy model à la Rota et al. (2016) here with unmarked and there with Jags and Nimble\nBias in occupancy estimate for a static occupancy model, with an interactive dashboard \u0026ndash; R and unmarked\nSpatial Animated map of wolf presence in France \u0026ndash; R\nDistribution map from an occupancy model, with Eurasian or Balkan lynx \u0026ndash; R, unmarked\nIntroduction to GIS and mapping in R using the sf package with slides \u0026ndash; R\nEpidemiological models and ODEs Dynamic models based on ODEs here and there \u0026ndash; Nimble\nExplore strategies of social distancing on epidemic w/ SIR model \u0026ndash; R\nZombie apocalypse and a Bayesian approach for fitting an epidemiological model to data \u0026ndash; R, OpenBUGS\nSocial networks Network analysis of my Twitter account \u0026ndash; R\nNetwork analysis of Star Wars The Attack of the Clones \u0026ndash; R\nAnimal social networks from capture-recapture data\nAnalysis of Bottlenose dolphin social network \u0026ndash; R\nScientific research is all about networking \u0026ndash; R\nDeep learning and species identification Piégeage photo et pipeline pour l\u0026rsquo;identification d\u0026rsquo;espèces via deep learning \u0026ndash; Python, R Bibliometric and textual analyses Content analysis of French articles in newspapers about wolf \u0026ndash; R\nBibliometric and textual analyses of the last decade of research in capture-recapture \u0026ndash; R\nQuick and dirty analyses of ISEC 2018 tweets \u0026ndash; R\nText mining of the ISEC 2020 abstracts \u0026ndash; R\nTidyverse Introduction to the tidyverse \u0026ndash; R\nSome notes I have taken on David Robinson\u0026rsquo;s screencasts, with R tidyverse tips and tricks \u0026ndash; R\nAnalyses of wolf depredation in France \u0026ndash; R\nVisualizing trends in French deaths \u0026ndash; R\nRandom stuff Decoding the hidden message in Perseverance parachute with R \u0026ndash; R\nVisualising the Metropolis algorithm on a simple 1D example \u0026ndash; R\nExport, clean up and tabulate questions and answers from a Slack work space with several channels into HTML tables \u0026ndash; R and json\nIllustration of the delta-method \u0026ndash; R\nCreating a hex sticker for the R2ucare package \u0026ndash; R\nSimulate data with Jags \u0026ndash; R, Jags\nFirst steps in Python \u0026ndash; Python\nBayesian Multivariate Autoregressive State-Space models - MARSS \u0026ndash; R, Jags\n","date":-62135596800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":-62135596800,"objectID":"fcef553317377ec10e2c25c685386d28","permalink":"https://oliviergimenez.github.io/coding/","publishdate":"0001-01-01T00:00:00Z","relpermalink":"/coding/","section":"","summary":"Below I provide pieces of code I have written over the years and shared through GitHub on reproducible science, capture-recapture, occupancy and epidemiological models, spatial stuff, social networks, deep learning, textual analyses, the tidyverse and some random stuff.","tags":null,"title":"Random pieces of code","type":"page"}] \ No newline at end of file diff --git a/index.xml b/index.xml index 9c956c7a0..0fd63b7a6 100644 --- a/index.xml +++ b/index.xml @@ -5,13 +5,61 @@ https://oliviergimenez.github.io/ Olivier Gimenez - Source Themes Academic (https://sourcethemes.com/academic/)en-us© Olivier Gimenez 2024Mon, 11 Dec 2023 00:00:00 +0000 + Source Themes Academic (https://sourcethemes.com/academic/)en-us© Olivier Gimenez 2024Fri, 23 Feb 2024 00:00:00 +0000 https://oliviergimenez.github.io/img/flyfishing.jpg Olivier Gimenez https://oliviergimenez.github.io/ + + Current lab members + https://oliviergimenez.github.io/people/current/ + Fri, 23 Feb 2024 00:00:00 +0000 + https://oliviergimenez.github.io/people/current/ + <br> +<hr> +<p><img src="https://oliviergimenez.github.io/img/adelaide.jpeg" width="150"> <strong>Célian Monchy</strong> (PhD 2022-2025) is working on the propagation of uncertainties related to the automatic processing of sensor data in different models in ecology. He is co-supervised by +<a href="https://marieetienne.github.io/" target="_blank" rel="noopener">Marie-Pierre Etienne</a>.</p> +<hr> +<p><img src="https://oliviergimenez.github.io/img/lea.jpg" width="130"> <strong>Léa Pautrel</strong> (PhD 2022-2025) is investigating the use of deep learning coupled with continuous-time models to exploit sensor data, in particular cam traps. She is co-supervised by +<a href="https://marieetienne.github.io/" target="_blank" rel="noopener">Marie-Pierre Etienne</a> and Sylvain Moulherat from +<a href="https://www.terroiko.fr/fr" target="_blank" rel="noopener">TerrOïko</a>.</p> +<hr> +<p><img src="https://oliviergimenez.github.io/img/maelis.jpeg" alt=""> +<strong>Maëlis Kervellec</strong> (PhD 2021-2024) is developing statistical models for the analysis of camera trap data with applications in large carnivores population ecology. This is joint work with colleagues from the +<a href="https://ofb.gouv.fr/" target="_blank" rel="noopener">French Office of Biodiversity</a></p> +<hr> +<p><img src="https://oliviergimenez.github.io/img/marwan.jpg" alt=""> +<strong>Marwan Naciri</strong> (PhD 2021-2024) is studying the demography of polar bears. He is co-supervised by Sarah Cubaynes and Jon Aars.</p> +<hr> +<p><img src="https://oliviergimenez.github.io/img/oksana.png" alt=""> +<strong>Oksana Grente</strong> (PhD 2018-2021, post-doc 2023-2024) works at evaluating the impact of lethal control on the French wolf population, with a special interest in social structure and livestock depredation. This is joint work with Sarah Bauduin, Christophe Duchamp and Nolwenn Hoguet-Drouet from +<a href="https://ofb.gouv.fr/" target="_blank" rel="noopener">French Office of Biodiversity</a>.</p> +<hr> +<p><img src="https://oliviergimenez.github.io/img/simon.jpeg" alt=""> +<strong>Simon Lacombe</strong> (PhD 2023-2026) is a quantitative ecologists who&rsquo;s determining the drivers of European otter recolonization in France using non-invasive sampling and statistical ecology. He is co-supervised by Sébastien Devillard.</p> +<hr> +<p><img src="https://oliviergimenez.github.io/img/javi.png" alt=""> +<strong>Javi Fernández-López</strong> (post-doc 2022-2023) is interested in conservation biology and phylogeography. He is modelling the distribution of mammals in Europe by integrating various sources of data. Check out his website +<a href="https://allthiswasfield.blogspot.com" target="_blank" rel="noopener">https://allthiswasfield.blogspot.com</a>.</p> +<hr> +<p><img src="https://oliviergimenez.github.io/img/monia.jpg" alt=""> +<strong>Mónia Nakamura Mercier Real</strong> (PhD 2019-2024) is investigating the impacts of anthropogenic factors on large carnivores persistence in human-dominated landscapes. She uses wolf in Spain and Portugal as a model species to address how different humanization levels can influence habitat selection and land use, and ultimately affect its demography, status and viability. The thesis is co-supervised with +<a href="https://scholar.google.es/citations?user=mWcf6MsAAAAJ&amp;hl=es" target="_blank" rel="noopener">José Vicente López-Bao</a> and +<a href="https://scholar.google.com/citations?user=2ulB-8QAAAAJ&amp;hl=en" target="_blank" rel="noopener">Pedro Monterroso</a>.</p> +<hr> +<p><img src="https://oliviergimenez.github.io/img/bea.jpg" alt=""> +<strong>Béatrice Chataigner</strong> (EPHE diploma 2020-2022) is working on locally-based monitoring programs in Kenya. Her research project aims at exploring the performances of current ecological monitoring program of Kenyan conservanciesin addressing their needs in terms of adaptive management at local level as well as ecosystem level. Béatrice works for IUCN in the +<a href="https://papaco.org/" target="_blank" rel="noopener">Program on African Protected Areas and Conservation</a>. Her work is jointly supervised by +<a href="https://institute.sandiegozoo.org/staff/nicholas-pilfold-phd" target="_blank" rel="noopener">Nicholas Pilfold</a> and +<a href="https://www.cefe.cnrs.fr/fr/actus/44-french/recherche/bc/bev/291-aurelien-besnard" target="_blank" rel="noopener">Aurélien Besnard</a>.</p> +<hr> +<p><img src="https://oliviergimenez.github.io/img/patricia.jpg" alt=""> <strong>Patrícia Rodrigues</strong> (post-doc 2021-2023) is interested in the social-ecological dynamics underpinning environmental problems in Africa. Her current project aims to characterize landscapes within Africa’s Great Green Wall in terms of their social-ecological properties and restoration initiatives. More +<a href="https://rodriguespatri.wixsite.com/home" target="_blank" rel="noopener">here</a>.</p> + + + Former lab members https://oliviergimenez.github.io/people/alumni/ diff --git a/people/current/index.html b/people/current/index.html index c566ec102..a241cf64c 100644 --- a/people/current/index.html +++ b/people/current/index.html @@ -27,7 +27,7 @@ - + @@ -274,13 +274,13 @@ - + - + - + @@ -572,7 +572,7 @@

    Current lab members

    - Jul 22, 2023 + Feb 23, 2024 @@ -614,7 +614,7 @@

    Current lab members


    -

    Adé Monchy (PhD 2022-2025) is working on the propagation of uncertainties related to the automatic processing of sensor data in different models in ecology. He is co-supervised by +

    Célian Monchy (PhD 2022-2025) is working on the propagation of uncertainties related to the automatic processing of sensor data in different models in ecology. He is co-supervised by Marie-Pierre Etienne.

    Léa Pautrel (PhD 2022-2025) is investigating the use of deep learning coupled with continuous-time models to exploit sensor data, in particular cam traps. She is co-supervised by @@ -956,7 +956,7 @@

    Statistics, ecology & social sciences