Table 2.

Performance Statistics for MCF7-Trained Candidate High Performance Classifiers

MIE Name	Classification Algorithm	Internal Accuracy	Holdout Accuracy	MIE Active Profiles	MIE Active Chemicals	Mean Null Accuracy	Empirical p-value
ADRA2A (+)	SVM_R	0.72	0.86	58	7	0.60	0.03
ALOX5 (−)	NB	0.73	0.55	51	5	0.61	0.01
AR (+)	NB	0.71	0.60	52	8	0.61	0.03
DRD2 (−)	SVM_R	0.68	0.54	118	14	0.58	0.03
ESR-1/2 (−)	MLP	0.89	0.92	68	5	0.69	0.00
ESR-1/2 (+)	SVM_L	0.85	0.79	145	12	0.64	0.00
FLT1/KDR (−)	MLP	0.75	0.69	122	10	0.66	0.02
HDAC (−)	SVM_L	0.82	0.78	174	10	0.67	0.00
HMGCR (−)	MLP	0.79	0.85	50	4	0.66	0.03
HRH1 (−)	MLP	0.71	0.61	110	14	0.61	0.01
JAK2 (−)	SVM_L	0.88	0.85	54	5	0.71	0.01
KCNH2 (−)	SVM_R	0.66	0.64	369	34	0.58	0.00
MAPK14 (−)	SVM_L	0.86	0.93	78	5	0.73	0.03
MET (−)	SVM_L	0.83	0.70	114	7	0.70	0.01
MTOR/PI3K (−)	SVM_R	0.90	0.88	204	12	0.70	0.00
NR3C1 (+)	SVM_R	0.73	0.68	100	10	0.60	0.01
PTGS-1/2 (−)	SVM_R	0.65	0.65	247	28	0.58	0.00
SLC22A6 (−)	KNN	0.70	0.64	55	6	0.58	0.02
TOP2A (−)	SVM_L	0.88	0.87	75	7	0.67	0.00
TUB (−)	SVM_L	0.94	0.90	104	8	0.59	0.00