The Environmental Performance Index, Visualized with R

The Environmental Performance Index (EPI) ranks countries on performance indicators for environmental public health and ecosystem vitality. Yale University hosts the EPI website, which was used to present the 2012 EPI Rankings to world leaders at the 2012 World Economic Forum at Davos.

The Environmental Performance Index (EPI) ranks countries on performance indicators for environmental public health and ecosystem vitality. Yale University hosts the EPI website, which was used to present the 2012 EPI Rankings to world leaders at the 2012 World Economic Forum at Davos. The Country Profiles section of the website allowed members to browse the performance characteristics of their countries, and see how they compared with other countries with similar environmental performance. (Switzerland was the 2012 overall leader.)

In the Visualizing and Interacting with Data Over the Web session at JSM 2012 this morning, Yale professor Jay Emerson revealed details of how this interactive website was created. The profile chart you see above is generated dynamically, using R grid graphics and the FastRWeb web framework for R.  The interactive “EPI Performance Score versus Trend Score” graphic was created using the gridSVG package. The country map and numeric tables were embedded using the googleVis package. And the “Countries with similar levels of performance” were generated using one of R’s many clustering algorithms.

The EPI website provides a great example of how R can be used to power an attractive, informative and useful interactive website.  And because all of these charts and analyses are generated dynamically from live data (rather than pre-rendered in advance), this same website will be trivial to update when the 2013 EPI performance scores are announced.

If you’d like more details of how these charts were created, check out Jay’s slides on grid graphics from the Introduction to Data Visualization and Analysis course at JSM. (There’s also a very useful section on parallel programming with the foreach and doSNOW packages.) Jay also provides some useful notes on setting up fastRWeb and Rserve on his blog.