Spatial Data Science with R, MSc course, 2019

Goals

The goals of this course are to get familiar with data science, with spatial data in data science problem, with techniques and tricks to using R for solving smaller problems, and with programming R to solve larger problems. The final assignment is one of:

• writing an R package for solving a particular spatial data science problem that passes all checks cleanly, and that contains one or more vignettes illustrating the problem(s) solved.
• develop a worked out spatial data science case study, wrapped as R markdown file, submitted in a fully reproducible manner

Motivation

Carrying out a proper scientific data analysis exercise involves not only understanding appropriate statistical methods, but also understanding the data, of the domain from which the data and the research question originate, and familiarity with tools that are needed or that still need to be developed to carry out the analysis. Ideally, the analysis should be carried out such that the report does not only give full insight in all details of the analysis, but also minimises the effort to reproduce or modify the complete analysis, and by that inviting to recreate the report itself, or modify it. The combinations of all these skills can be called data science.

Spatial data science is a flavor of data science where spatial locations play a role, not only to hold records together (which IDs also could do), but also play an explicit role in the analysis, e.g. to compute and analyse distances, consider spatial autocorrelation, or compute spatial relationships between features with different geometry types, such as distances from points to linestrings. Quite often, temporal considerations play a role too when considering spatial properties.

R is a language and environment for statistical computing that plays a large role in data science, because it is open source, it is easily extendible, and it used a lot by non-computer programmers to carry out programming tasks.

Course organisation

The course will be organized along the following topics:

1. (Apr 8) R and its package ecosystem; CRAN; other ecosystems: bioconductor, github; flavors of R, IDEs, rstudio questions
2. (Apr 15) R basics: objects, vectors, arrays, lists, NULL, attributes; data.frames
3. (Apr 22) (no class) (Apr 25) (no class)
4. (Apr 29) R models, formula, lm, predict, plot, summary, subset, NA, factor
5. (May 3) R spatial basics: history, packages, sys reqs, simple features
6. (May 6) Vector geometric operations, DE-9-IM
7. (May 13) Attributes, attribute-to-geometry relations
8. (May 20) Raster, vector-raster conversion, arrays
9. (May 27) R plotting: base plot, grid, lattice, ggplot2
10. (Jun 3) Tidyverse basics: tidy data, tidy tools manifesto, dplyr, pipes
11. (Jun 17) R packages; functions, methods, S3, S4, R6
12. (Jun 24) R looping: for, apply, lapply, purrr
13. (Jul 1) Extending R: foreign language interfaces; R markdown;

References

• [JC] John Chambers, 2016. Extending R. CRC Press.

• [AR] Hadley Wickham, Advanced R. http://adv-r.had.co.nz/

• [RP] Hadley Wickham, R Packages. http://r-pkgs.had.co.nz/

• [RDS] Hadley Wickham and Gareth Grolemund, R for Data Science, http://r4ds.had.co.nz/

• [ASDAR] Roger Bivand, Edzer Pebesma, Virgilio Gomez-Rubio, Applied Spatial Data Analysis with R. Springer, NY.; http://www.asdar-book.org/

• [IR] R Core Team, 2017, An Introduction to R