Figure 8From: Win percentage: a novel measure for assessing the suitability of machine classifiers for biological problemsSelection pressure as a function of N. Each plot shows the probability mass function for feature sets ranked by estimated performance. We consider a total of M= 100 unique feature sets each with a corresponding estimated performance. The five line plots show the probability that each of the M feature sets will be selected as best in a random selection of N sets, with replacement. As N increases, the feature sets with higher performance (lower rank) are more likely to be selected.Back to article page