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

Table 5 Evaluation results of feature sets of the drug–target interaction using machine and ensemble algorithms according to precision, recall, F-score, and accuracy

Feature set	Prediction algorithms	Precision	Recall	F-score	Accuracy
Feature set [1]	SVM	0.995	0.995	0.995	0.996
	RF	0.9996	0.9996	0.9996	0.9997
	AB	0.9998	0.9998	0.9998	0.9999
	XG	0.9994	0.9995	0.9995	0.9996
	Light	0.9997	0.9997	0.9997	0.9998
Feature set [2]	SVM	0.9992	0.9992	0.9992	0.9991
	RF	0.9996	0.9996	0.9996	0.9996
	AB	0.9998	0.9998	0.9998	0.9998
	XG	0.9995	0.9995	0.9995	0.9996
	Light	0.9996	0.9996	0.9996	0.9997
Feature set [3]	SVM	0.992	0.992	0.992	0.992
	RF	0.9993	0.9993	0.9993	0.9992
	AB	0.9993	0.9993	0.9993	0.999
	XG	0.999	0.999	0.999	0.9988
	Light	0.9989	0.9989	0.9989	0.9987
Feature set [4]	SVM	0.951	0.948	0.948	0.942
	RF	0.999	0.999	0.999	0.9989
	AB	0.9992	0.9992	0.9992	0.9989
	XG	0.999	0.999	0.999	0.9987
	Light	0.9988	0.9988	0.9988	0.998
All Feature set	SVM	0.993	0.993	0.993	0.994
	RF	0.9992	0.9992	0.9992	0.9993
	AB	0.9993	0.9993	0.9993	0.9993
	XG	0.998	0.998	0.998	0.998
	Light	0.9991	0.9991	0.9991	0.999

ISSN: 1754-1611