Table 2.

The performance evaluation of the feature subsets selected by different filter feature selection methods

	IG		CHI		ReliefF		CFS
Peformance	200	300	200	300	200	300	200	300
Specificity	0.759	0.759	0.755	0.76	0.751	0.772	0.733	0.746
Sensitivity	0.704	0.718	0.707	0.704	0.718	0.743	0.727	0.731
F-measure	0.73	0.738	0.73	0.731	0.734	0.757	0.73	0.738
ROC score	0.74	0.745	0.738	0.741	0.74	0.762	0.731	0.741

Based on the feature subsets selected by four filter feature selection methods, we respectively build SVM classifiers and compare the performance of these classifiers.