GGally: A Plot Matrix for All Variable Types

• Activity Number: 80
• Type: Contributed
• Date/Time: Sunday, August 1, 2010 : 4:00 PM to 5:50 PM
• Sponsor: Section on Statistical Graphics
• Author(s): Barret E. Schloerke*+ and Dianne Cook and Hadley Wickham and Heike Hofmann
• Companies: Iowa State University and Iowa State University and Rice University and Iowa State University
• Keywords: scatterplot matrix ; multivariate data ; data visualization ; composite displays ; ggplot2

Abstract

Scatterplot matrices are limited for viewing multivariate data because they only allow pairwise plots of real-valued variables. We've re-designed the scatterplot matrix to handle other variable types, and implemented this in a new R package called Gally. GGally accepts common X vs. Y combinations of data types and makes an intelligent guess at the best type of pairwise plot to use, for example, mosaic plot, side-by-side boxplots, or scatterplot. The upper triangle, and lower triangle, may contain different plot types for the same pair of variables. Each sub plot can be retrieved, altered, and/or completely replaced by a custom-produced display. The software is built upon Hadley Wickham's ggplot2 package, and builds on ideas contained in Jay Emersons gpairs plot in the YaleToolkit package. Future work will expand the offerings of composite types of displays.