Table 2.

Comparison of classifier performance on human datasets

Classifier	HumDiv				HumVar
	FPR			acc	FPR
	0.05	0.10	0.20		0.05	0.10	0.20
	TPR				TPR			acc
Deep Forest	0.950	0.986	0.996	0.951	0.584	0.749	0.876	0.842
Random Forest	0.947	0.986	0.997	0.940	0.563	0.733	0.868	0.830
Neural Network	0.916	0.972	0.993	0.939	0.548	0.716	0.857	0.827
Gaussian SVM	0.916	0.975	0.995	0.940	0.551	0.719	0.857	0.829
Polynomial SVM	0.917	0.973	0.995	0.940	0.549	0.716	0.854	0.828
Logistic Regression	0.895	0.961	0.992	0.931	0.484	0.666	0.831	0.814
Linear SVM	0.897	0.961	0.992	0.933	0.483	0.667	0.831	0.815
Boosted Gaussian NB	0.850	0.936	0.977	0.805	0.445	0.650	0.822	0.812
Gaussian NB	0.794	0.928	0.978	0.805	0.341	0.568	0.813	0.812
PolyPhen‐2	0.78	0.89	0.96	0.89	0.53	0.68	0.83	0.81

TPRs (true‐positive rates) corresponding to a given FPR (false‐positive rate) are provided. The values of the accuracy metric (acc) are given for all classifiers ordered by decreasing of AUC metric (see Figure 1 and Figure S2). Cells with AUC or acc values no less than corresponding PolyPhen‐2 values are filled in yellow, while cells with AUC or acc values smaller than corresponding PolyPhen‐2 values are colored light blue. In each column, the maximal value is in bold.