Figure 3From: Win percentage: a novel measure for assessing the suitability of machine classifiers for biological problemsWin percentage as a function of N. Win percentage for the three candidate classifiers defined in Equation 1. The dotted lines correspond to the exact win percentage for each classifier as a function of the number of randomly selected feature sets, N. The solid lines show the 95% confidence interval for the sampled estimate of win percentage using M= 10,000 randomly drawn feature sets from the joint density displayed in Figure 2.Back to article page