Table 1.

Performance comparison of six classical classification models on random selection method, NCPIS and NDTISE.

Metric	Negative DTIs	NB	kNN	L1-R	L2-R	RF	SVM
Precision	Random	0.338	0.458	0.786	0.787	0.529	0.700
	NCPIS	0.361	0.716	0.823	0.837	0.847	0.969
	NDTISE	0.422	0.759	0.877	0.842	0.840	0.965
Recall	Random	0.376	0.306	0.622	0.631	0.306	0.261
	NCPIS	0.560	0.882	0.749	0.773	0.824	0.883
	NDTISE	0.625	0.897	0.775	0.817	0.821	0.876
F-measure	Random	0.356	0.367	0.694	0.700	0.388	0.380
	NCPIS	0.439	0.790	0.784	0.804	0.835	0.924
	NDTISE	0.504	0.822	0.823	0.829	0.830	0.918
AUC	Random	0.622	0.593	0.879	0.873	0.694	0.705
	NCPIS	0.672	0.904	0.917	0.920	0.954	0.942
	NDTISE	0.752	0.928	0.934	0.939	0.948	0.941