Collaborative Data Analysis with Rcloud

• Activity Number: 472
• Type: Invited
• Date/Time: Wednesday, August 6, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Graphics
• Author(s): Carlos E. Scheidegger*+
• Companies: AT&T Labs

Abstract

Data analysis tasks are too complicated today for one single statistician, as can be witnessed by the proliferation of larger teams of statisticians. It is imperative, then, that we support the task of collaboratively exploring and understanding a data analysis problem. RCloud is a web-based environment for developing and sharing data analysis scripts written in R. It provides a low-friction path from developing scripts to sharing, publishing and turning scripts into web services. It provides transparent, integrated version control and social features such as searching for other scripts, commenting, "forking" and "cloning". This way, institutional memory is captured much more effectively than with a traditional toolchain. In addition, because RCloud works over a web browser, it can leverage libraries for graphics such as d3, dc.js and leaflet, and offer integrated rich interaction capabilities beyond what's possible in typical R deployments. I will present and discuss RCloud's features in the context of the current needs of applied statisticians and data analysts, and speculate on where we should go next.