Odum Institute: Modular design and automated testing in R
Modular design and automated testing in R
This course will be split over 2 days and offered via Zoom. Attendance is required as the course will not be recorded.
Research is relying on increasingly complex code. However, most academics have not received formal training in software development. This course will introduce participants to some of the best practices of professional software development. Participants will learn to structure their R code into reusable functions and store those functions as separate scripts and/or packages. Participants will learn to write automated tests for their functions, to help confirm that results are as expected and to ensure that future changes to code do not cause unexpected changes to results.
We will also discuss how to create reproducible “environments” that record package versions in use to help avoid package version incompatibilities and ensure that results can be replicated even months or years after the original analysis.
Modular design is best paired with a version control system. We will discuss how modular design practices integrate with Git and Github, and how to set up “continuous integration” so that automated tests are run each time code is changed.
The course will primarily consist of a hands-on exercise to restructure a single long analysis file into functions and automated tests.