Reproducible research

I spend a lot of time attempting to reproduce historical work (including my own) and strongly feel that all research, especially that paid for with public money, should be 100% reproducible where possible. This is especially important in natural resource management where research advice is often subject to intense public scrutiny and the integrity of the decision making process is paramount.

In recent years I have experimented with various workflows for integrating data analysis and version control into my reporting. My ultimate goal is to have a workflow that is fully reproducible, open source, and takes care of the ‘last mile’. The final output should be the pre-print version of the report or academic manuscript, ready for submission.

I currently use the R language, coupled with R-Studio and the tidyverse suite of packages, Github for version control, Zotero for managing references and Packrat for managing package dependencies. Quite a bit of inspiration for my current workflow came from this blog post.

My first attempt can be found here.

My second attempt, which involves a lot more data analysis, can be found here.