Identification and Analysis of Driver Missense Mutations Using Rotation Forest with Feature Selection

Table 3

Performance of predicting on three test datasets (TP53, EGFR, and Cosmic2plus).

Method	Test set	Accuracy	Recall	Precision	F-measure	MCC
mRMR-RF	TP53 + neutral	88.86	100	62.4	76.85	0.734
	EGFR + neutral	86.68	100	15.88	27.41	0.3702
	Cosmic2plus + neutral	85.3	81.04	59.26	68.46	0.6041
DX-LibSVM	TP53 + neutral	83.93	100	53.48	69.69	0.6553
	EGFR + neutral	80.78	100	11.56	20.73	0.3047
	Cosmic2plus + neutral	81.51	86.52	51.83	64.83	0.5655
DX-SVMLight	TP53 + neutral	88.31	100	61.25	75.97	0.7243
	EGFR + neutral	86.02	100	15.23	26.44	0.3612
	Cosmic2plus + neutral	85.42	84.46	59.08	69.53	0.6199
DX-RF	TP53 + neutral	89.28	100	63.28	77.51	0.7414
	EGFR + neutral	87.18	100	16.39	28.16	0.3772
	Cosmic2plus + neutral	85.53	80.14	59.91	68.56	0.6048