TABLE 3.

Performance comparison of different classifiers.

Datasets	Classifier	AUC	AUPR	Recall	Precision	F1-score
NPInter2.0 (5:5)	SVM	0.902	0.840	0.979	0.847	0.908
	XGBoost	0.640	0.637	0.285	0.980	0.442
	GBDT	0.556	0.553	0.117	0.957	0.209
	Random forest	0.937	0.906	0.946	0.930	0.938
	KNN	0.935	0.898	0.963	0.913	0.937
	LR	0.950	0.899	0.919	0.886	0.902
NPInter2.0 (4:6)	SVM	0.879	0.814	0.958	0.827	0.887
	XGBoost	0.655	0.652	0.316	0.981	0.479
	GBDT	0.581	0.577	0.174	0.941	0.294
	Random forest	0.928	0.894	0.935	0.922	0.928
	KNN	0.931	0.862	0.955	0.911	0.932
	LR	0.944	0.893	0.920	0.901	0.899
NPInter3.0_H (5:5)	SVM	0.816	0.763	0.792	0.831	0.811
	XGBoost	0.640	0.637	0.285	0.980	0.442
	GBDT	0.642	0.617	0.246	0.976	0.394
	Random forest	0.830	0.808	0.710	0.934	0.807
	KNN	0.803	0.772	0.698	0.889	0.782
	LR	0.913	0.909	0.587	0.911	0.714
NPInter3.0_H (4:6)	SVM	0.829	0.807	0.709	0.933	0.805
	XGBoost	0.655	0.654	0.314	0.990	0.442
	GBDT	0.619	0.604	0.240	0.996	0.386
	Random forest	0.872	0.855	0.781	0.954	0.859
	KNN	0.855	0.825	0.793	0.916	0.843
	LR	0.923	0.935	0.682	0.927	0.786
NPInter3.0_M (5:5)	SVM	0.825	0.741	0.997	0.741	0.850
	XGBoost	0.652	0.652	0.305	0.995	0.468
	GBDT	0.617	0.605	0.226	0.996	0.352
	Random forest	0.904	0.887	0.872	0.932	0.901
	KNN	0.915	0.876	0.924	0.907	0.915
	LR	0.866	0.791	0.945	0.747	0.835
NPInter3.0_M (4:6)	SVM	0.733	0.653	0.976	0.656	0.785
	XGBoost	0.644	0.639	0.300	0.962	0.458
	GBDT	0.603	0.582	0.218	0.994	0.327
	Random forest	0.870	0.837	0.820	0.911	0.863
	KNN	0.883	0.835	0.895	0.875	0.884
	LR	0.803	0.679	0.741	0.709	0.725
RPI2241 (5:5)	SVM	0.630	0.576	0.908	0.584	0.711
	XGBoost	0.606	0.601	0.212	0.991	0.350
	GBDT	0.616	0.589	0.195	0.982	0.356
	Random forest	0.937	0.893	0.985	0.898	0.940
	KNN	0.725	0.668	0.679	0.748	0.711
	LR	0.757	0.657	0.634	0.695	0.663
RPI2241 (4:6)	SVM	0.641	0.582	0.912	0.615	0.735
	XGBoost	0.625	0.613	0.241	0.971	0.416
	GBDT	0.626	0.612	0.253	0.964	0.375
	Random forest	0.914	0.905	0.981	0.892	0.958
	KNN	0.752	0.647	0.684	0.751	0.815
	LR	0.775	0.689	0.684	0.705	0.694

Bold values are the best performance of each task.