Facilitating Reproducible Research with R, Markdown, and Git

• Activity Number: 171
• Type: Topic Contributed
• Date/Time: Monday, August 10, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Consulting
• Author(s): Karthik Ram*
• Companies: UC Berkeley
• Keywords: reproducibility

Abstract

A vast majority of scientific studies published today remain irreproducible despite reproducibility being widely recognized as the hallmark of good scientific practice. Over the past two decades various tools have emerged to make computational reproducibility more accessible. Yet many of these seemingly popular tools have failed to gain traction. For example, `Sweave` in R has allowed for easy integration of code and `LaTeX` markup. However, the steep learning curve associated with Latex has proved to be a barrier for many researchers.

More recent developments with tools such as `knitr` and integration with IDEs like RStudio and cloud services like GitHub have made it exceedingly easy for researchers to write simple markup. Markdown, in particular, is a lightweight markup that is not only easy to write but also easy to read even in an unrendered state. Further, related tools such as those developed by the rOpenSci project make it easy to read web-based data directly into R, or deposit existing data into persistent repositories, parse results, and share rendered and raw documents on code repositories such as GitHub.

In this talk I'll provide both an overview and live demos of the