Figure 1
From: Using beta binomials to estimate classification uncertainty for ensemble models
Distribution of classification errors for the logP2 data set. Observed logP values for compounds in the data set were binned at intervals of 0.1 and counts within each bin are represented by the blue curve. Data for compounds with logP values below -4 are omitted for clarity, as are those above 8. The distribution of logP values for compounds misclassified by model logP2-1 is shown in red. Note that the scale for error counts is given at the right of the figure and is magnified by a factor of 2.