Assessing Confidence in Predictions.wmv
In this presentation, Philip Judson discusses assessing confidence in predictions. To measure performance using Cooper statistics (e.g. concordance) you need binary predictions – positive or negative. For programs like Derek and Meteor you have to make an arbitrary choice of threshold, such as classing all predictions with a likelihood of plausible or greater as positive, and all others as negative. This presentation will describe a new metric, Veracity.
Presented by Philip Judson as part of the 2013 virtual ICGM series; 22nd May 2013.