Table 2.

Average AUC and AUPR of different learning models on different feature sets, evaluated by cross validation. The standard deviation of every AUC and AUPR is less than 0.02.

Selection:	None		Chi-squared filter		CFS		Wrapper for AUC
Learning Models	AUC	AUPR	AUC	AUPR	AUC	AUPR	AUC	AUPR
Logistic regression	0.645	0.177	0.709	0.209	0.734	0.243	0.759	0.244
Cox	0.621	0.173	0.707	0.208	0.735	0.243	0.755	0.239
Naïve Bayes	0.689	0.194	0.706	0.226	0.726	0.232	0.745	0.240
CART	0.653	0.171	0.645	0.173	0.651	0.173	0.667	0.185
Random forest	0.696	0.203	0.708	0.210	0.666	0.174	0.757	0.235