Table 2.

Statistical metrics for the best algorithm of each cluster of method and their combined regression model, both the “Full Regression” and the stepwise-optimized regression model (rf + svmPoly + pda) for both training and testing set.

	C5.0		pda		plr		rf		svmPoly		Full Regression		rf + svmPoly + pda
	Train	Test	Train	Test	Train	Test	Train	Test	Train	Test	Train	Test	Train	Test
AUROC	0.88	0.83	0.85	0.84	0.83	0.85	0.93	0.83	0.89	0.83	0.91	0.91	0.91	0.91
Accuracy	0.88	0.91	0.85	0.88	0.83	0.85	0.93	0.90	0.89	0.90	0.94	0.95	0.94	0.95
Sensitivity	0.78	0.68	0.86	0.76	0.82	0.84	0.87	0.71	0.80	0.68	0.98	0.98	0.98	0.98
Specificity	0.98	0.98	0.84	0.91	0.85	0.85	0.98	0.96	0.98	0.97	0.84	0.85	0.84	0.85
PPV	0.98	0.90	0.84	0.73	0.84	0.64	0.98	0.84	0.97	0.87	0.95	0.95	0.95	0.95
NPV	0.81	0.91	0.85	0.93	0.82	0.95	0.89	0.91	0.83	0.91	0.91	0.94	0.91	0.94
FPR	0.22	0.32	0.14	0.24	0.18	0.16	0.13	0.29	0.20	0.32	0.02	0.02	0.02	0.02
FNR	0.02	0.02	0.16	0.09	0.15	0.15	0.02	0.04	0.02	0.03	0.16	0.15	0.16	0.15
F1	0.86	0.78	0.85	0.74	0.83	0.73	0.92	0.77	0.88	0.76	0.96	0.97	0.96	0.97

PCA: dataset upon Principal Component Analysis; PCAUp: dataset upon Principal Component Analysis and up-scaling of the minor class; PCADown: dataset upon Principal Component Analysis and down-sampling of the major class; Scaled: dataset upon z-score calculation; ScaledUp: dataset upon z-score calculation and up-sampling of the minor class; ScaledDown: dataset upon z-score calculation and down-sampling of the major class.