The jury consisting of Nils B Weidmann (University of Konstanz), Hanne Fjelde (Uppsala University) and Michael D Ward (Duke University) has awarded the fifth JPR Best Visualization Award to Michael Colaresi (University of Pittsburgh) and Zuhaib Mahmood (Michigan State University). The prize-winning article, titled ‘Do the robot: Lessons from machine learning to improve conflict forecasting’, was published in the May special issue on forecasting in peace research, Journal of Peace Research 54(2): 193–214. While the article makes extensive use of graphics to illustrate procedures, data and results, the jury was particularly impressed by the introduction of a new visual tool to assess the predictive accuracy of quantitative models. These ‘model criticism plots’ are part of an iterative cycle of model construction, evaluation and improvement, and help users identify observations where the difference between predicted and observed outcomes is particularly pronounced. In doing so, Colaresi and Mahmood’s graphs take the idea of separation plots (Greenhill, Ward & Sacks, 2011) one step further, and add an additional dimension to jointly evaluate absolute values, but also their rank in the overall sample.
The award is USD 500.