Roadmap

[robitalec]

• Consider what our general families of functions are and potentially rename functions. How can we rename these functions to form a more coherent set of functions, where we have a category or general group (underscore) specific type or qualifier. I think the extract_distance, extract_pt makes sense, but I think others could use work.

  • prepare_
  • calc_
  • rss_
  • speed_
  • extract_
  • model_
  • plot_
  • predict_
  • resample_

• Are there functions we don’t need? Are there situations where two functions could be combined by using an argument instead? Eg. plot_boxplot and plot_boxplot_horiz

• ?Adjust extract_layers to take arguments “point sample” “distance sample” “buffer sample” (https://github.com/robitalec/targets-issa/blob/main/R/extract_layers.R). Eg. point_sample = list(elevation, land_cover) which internally passes elevation raster and land_cover raster to extract_pt and names columns “point_elevation”. Eg. distance_sample = list(road) which internally passes road feature to extract_distance and names columns “distance_road”

• ?Manage if branched targets are NULL in functions (https://github.com/robitalec/targets-issa/blob/main/R/calc_distribution_parameters.R#L5C1-L5C32)

• Functions that users should adapt to their own needs. (related, can these be generalized?)

  • Plot functions eg. https://github.com/robitalec/targets-issa/blob/main/R/plot_distributions.R
  • Model functions eg. https://github.com/robitalec/targets-issa/blob/main/R/prepare_model.R, https://github.com/robitalec/targets-issa/blob/main/R/model_forest_bin.R
  • Predict eg. https://github.com/robitalec/targets-issa/blob/main/R/predict_h1_forest.R

• Reconsider how models are structured in the targets workflow. Should it be a function? Or should it just be inline command for a target?

• Wrap all of the plots and outputs into a vignette called “results” or “manuscript template”.

• Try to parse the model formulas into the manuscript document (see https://cran.r-project.org/web/packages/equatiomatic/index.html)

• Use R/variables.R to provide specific place to go to set core “settings” eg. all of these lines (https://github.com/robitalec/targets-issa/blob/main/_targets.R#L35-L58)

• Convert plot_theme() from a function to a variable in variables.R, I think a function is less intuitive to new users (https://github.com/robitalec/targets-issa/blob/main/R/plot_distributions.R#L8)

• Document how targets:all works (https://github.com/robitalec/targets-issa/blob/main/_targets.R#L275-L277)

• Document what each step is doing to highlight what kinds of things need to happen at this stage, to help users adapt to their contexts (https://github.com/robitalec/targets-issa/blob/main/R/prepare_model.R)

• Document each target and group of targets with inline comments

• Generally, try to point to external resources eg. iSSA book or papers or amt docs etc throughout the documentation

• Note to cite the targets-iSSA template in your own work if you worked from it

• Does this also need to retain the iteration (splitBy) in the output? (https://github.com/robitalec/targets-issa/blob/main/R/model_check.R)

• Replace raster with terra? Use geotargets? Note if you use geotargets/terra, you can’t return ggplot objects, we need to return saved figures or use base plotting

• Add roxygen documentation for all functions

• Make a pkgdown website

• Make a vignette for adapting the targets template and its functions to your own study

• Fix/flex colnames in make_track target #2

• Fix/flexible extract layers function #3

• Fix/match function and file names #4

• Fix/args not used #5

• Document system requirements from packages eg. gpm