An ensemble-based drug–target interaction prediction approach using multiple feature information with data balancing

Table 6 Record area under the curve (AUC), mean square error, and MCC are achieved by different techniques

Feature set	Prediction algorithms	AUC	Mean square error	MCC
Feature set [1]	SVM	0.9954	0.0047	0.99
	RF	0.9996	0.00038	0.9993
	AB	0.9998	0.00023	0.9996
	XG	0.9995	0.0005	0.9991
	Light	0.9997	0.0003	0.9996
Feature set [2]	SVM	0.981	0.0008	0.998
	RF	0.9996	0.00035	0.9993
	AB	0.9998	0.00015	0.9997
	XG	0.9995	0.0004	0.9994
	Light	0.9996	0.0004	0.9991
Feature set [3]	SVM	0.976	0.0082	0.984
	RF	0.9993	0.0007	0.9986
	AB	0.9993	0.0007	0.9986
	XG	0.999	0.0009	0.9982
	Light	0.9989	0.001	0.9979
Feature set [4]	SVM	0.949	0.051	0.8997
	RF	0.999	0.0009	0.998
	AB	0.9992	0.0008	0.998
	XG	0.999	0.0009	0.998
	Light	0.9988	0.001	0.997
All feature sets	SVM	0.993	0.007	0.986
	RF	0.9992	0.0008	0.999
	AB	0.9993	0.00067	0.999
	XG	0.998	0.0018	0.996
	Light	0.9991	0.00085	0.998

ISSN: 1754-1611