TABLE 6.

Comparison between the results of different datasets on four classifiers.

Data category	Testing accuracy				Testing balanced accuracy
	SVM	RF	KNN	Proposed	SVM	RF	KNN	Proposed
Gene	0.9878	0.9918	0.9878	0.9918	0.8995	0.9707	0.9619	0.9481
PMA50	0.9743	0.9869	0.9824	0.9910	0.8831	0.8980	0.9736	0.9342
Integrative dataset	0.9865	0.9902	0.9914	0.9951	0.9413	0.9208	0.9408	0.9731

For the gene expression, the proposed method obtains the highest accuracy, but the balanced accuracy is highest in RF. For the PMA50, the proposed method obtains the best accuracy. For the integrative dataset, the proposed method obtains the highest accuracy and balanced accuracy, which illustrates that the integrative dataset contains more useful information after feature selection. The bold values are the best results.