Online Program

Reporting Results with R, R Markdown, and Shiny (302968)

Keywords: R, reporting, reproducibility, literate programming, interactive, shiny, rmarkdown, packrat

A data scientist is not finished when his or her result is calculated. A good scientist must report the results to a skeptical (or worse, uninformed) audience. This talk examines a tool chain for easy, enhanced reporting with R. The Rmarkdown package lets users combine human readable text with machine readable code to produce literate programs that are easy to understand and easy to reproduce. Users can turn the program into an HTML, PDF, or Microsoft Word report with one click of a button. The R code chunks in the report can be replaced by the results of the code chunk, at the user’s discretion. The Packrat package extends this workflow to make R analyses more reproducible. Packrat creates a system for recording and sharing the packages used in an analysis. Colleagues can use this system to reproduce the analysis in its original computing environment. The Shiny package completes the workflow. Shiny maps R code to controls embedded in the report, which lets your audience change the parameters of an analysis and see the results in real time. Together, these tools quickly create reports that are easy to reproduce and easy to understand.