I gave a talk today at R-Ladies Dallas via Zoom. The meetup was recorded and a video is now available here. The slides are here. TL;DR you should not do the following things for reproducibility within and across research projects:
- Not working in an R project
- Not using Git (version control)
- Not using renv package
- Not having a consistent directory structure
- Messing with your raw data
- Not having a set of ordered, modular R scripts
- Copy-pasting code
- Not commenting properly
- Not writing/deploying tests
- Not automating figures/tables
The time you spend to make things reproducible will come back to you by * not* being wasted on mending broken things later. Ping me via email or Twitter if you have additional things that you want to add!