The winners of this year's ggplot2 case study competition have been announced. I was honoured to be asked to be a judge of the competition this year, but it was a difficult job with so many excellent entries. In the end, the judging panel (which included Heike Hoffman and Hadley Wickham and me) selected three entries which each demonstrated the ability to convey meaning in complex data through elegant visualizations, while using some of the more powerful features of the ggplot2 library in R.
First prize goes to David Kahle of Rice University, who integrated public crime statistics and Google Maps to create this "violent crime weather map" for the city of Houston, Texas:
While the underlying data are quite complex, this is an immediately understandable and relatable representation of the parts of Houston where violent crime is most prevalent. David wins a 32 Gb iPod Touch, donated by Revolution Analytics.
There were also two runners-up in the competition, appearing after the jump.
Finalist Claudia Beleites (Technische Universität Dresden / Università degli Studi di Trieste) created a visualization to aid with classification of brain tumors using Raman spectroscopy. After decomposing the spectra using linear discriminant analysis, the three tumor types can be identified (except in the overlapping areas) using 3-d histograms.
Claudia wins a Use R! book of her choice, donated by Springer.
Finalist Michael Lavine (UMass Amherst) also looked at data from brains, but this time to identify regions that are "hyperexcitable", where excessing electrical activity can lead to epileptic seizures. Working from a series of photographs of the living brain surface subjected to electrical probes (and eliminating effects due to heartbeat and respiration), he created these phase plots of electrical intensity versus rate of change over time (the color) for four regions of the brain under two different kinds of EEG stimulus.
Congratulations to all of the entrants in the competition for an impressively high standard of data visualization. Check out all the entries (each with associated R code) at the link below.
ggplot2 wiki: ggplot2 case studies 2010