Table 1.

The prediction performance by five classification approaches, C4.5 decision tree, Bayes network, SVM with the RBF kernel, SMO and nearest neighbor

Strategy	Algorithm	Sn	Sp	Ac	MCC	AUC
10FCV Training	C4.5	0.8766	0.7832	0.8371	0.6646	0.832
	Bayes net	0.6594	0.8034	0.7203	0.4585	0.807
	RBF	0.9463	0.7613	0.8681	0.7315	0.854
	SMO	0.9474	0.8783	0.9182	0.8321	0.913
	NN	0.9177	0.8128	0.8734	0.7395	0.865
Testing	C4.5	0.7667	0.8627	0.8108	0.6280	0.811
	Bayes net	0.6333	0.8039	0.7117	0.4398	0.764
	RBF	0.8500	0.8235	0.8378	0.6735	0.837
	SMO	0.8667	0.9020	0.8829	0.7664	0.884
	NN	0.8667	0.8039	0.8378	0.6730	0.835
10FCV All	C4.5	0.8759	0.8020	0.8445	0.6809	0.841
	Bayes net	0.6877	0.7905	0.7314	0.4730	0.815
	RBF	0.9358	0.7746	0.8672	0.7290	0.855
	SMO	0.9422	0.8887	0.9195	0.8349	0.915
	NN	0.9112	0.8165	0.8709	0.7349	0.864

These algorithms were evaluated using the 10FCV on the Training dataset, the Testing dataset and the All dataset.