Identification of Estrogen Receptor α Antagonists from Natural Products via In Vitro and In Silico Approaches

Table 2

Performance of Bayesian and recursive partitioning models and their 5-fold cross-validation results.

Model	TP	FN	FP	TN	SE	SP	MCC	Q^+	Q^−
NB-a	1091	35	223	5206	0.969	0.959	0.874	0.830	0.993
NB-b	1119	8	21	5408	0.993	0.996	0.985	0.982	0.999
NB-c	749	316	455	5037	0.704	0.917	0.591	0.622	0.941
NB-d	1054	11	75	5416	0.990	0.986	0.953	0.933	0.998
RP-a	1113	13	51	5379	0.988	0.991	0.966	0.956	0.998
RP-b	1111	15	49	5381	0.987	0.991	0.966	0.958	0.997
RP-c	1007	57	177	5314	0.946	0.968	0.876	0.850	0.989
RP-d	1022	42	95	5396	0.960	0.983	0.925	0.915	0.992

Note: NB-a: NB model with MOE 2D descriptors; NB-b: NB model with MOE 2D descriptors + ECFP_6; NB-c: NB model with DS 2D descriptors; NB-d: NB model with DS 2D descriptors + ECFP_6; RP-a: RP model with MOE 2D descriptors; RP-b: RP model with MOE 2D descriptors + ECFP_6; RP-c: RP model with DS 2D descriptors; RP-d: RP model with DS 2D descriptors + ECFP_6.