Paper presented at the Computer Applications and Quantitative Methods in Archaeology (CAA) conference, Tübingen, 2018.
From its origins as a statistical programming language, R and the R ecosystem has become the ‘Swiss Army knife’ of data science. There is a package for seemingly every quantitative computing task you can throw at it, whilst tools such as RMarkdown and Shiny extend its use beyond pure analysis. This raises the question: is there anything R can’t do? In this talk I offer my personal reflections on an attempt to produce a doctoral thesis entirely in R – from the initial stages of research to the preparation of a final manuscript. This experiment entailed using R for a number of tasks where it was not the obvious first choice: as a tool for field data collection; as a GIS; for constructing agent-based models; for collaborating with colleagues using R Notebooks and Shiny apps; and for preparing for submission an R Markdown document and accompanying git repository. The experiment was a success in that I found that R was a robust alternative for most of these tasks. Moreover, distilling the research process into a single ‘package’ offered a number of advantages over a more intuitive, fragmentary approach using different softwares. However, rather than preaching entirely to the converted, I will highlight some areas where relying solely on R was unnecessarily challenging, and where it is perhaps better used to supplement to other tools, or avoided altogether – at least in its current state.