Prediction and Screening Model for Products Based on Fusion Regression and XGBoost Classification

Table 6

The prediction accuracy of another five prediction methods and the accuracy comparison with XGBoost.

Prediction methods	ADMET properties	Prediction accuracy (%)	Accuracy improvement (%)
SVM	Caco-2	86.8	7.2
	CYP3A4	87.6	7.8
	hERG	86.3	3.1
	HOB	87.3	1.6
	MN	85.8	10.4
RF	Caco-2	93.2	0.8
	CYP3A4	94.2	1.5
	hERG	88.1	1.3
	HOB	88.4	0.5
	MN	94.9	1.3
LDA	Caco-2	88.5	5.5
	CYP3A4	93.2	2.5
	hERG	86.1	3.3
	HOB	85.3	3.6
	MN	87.6	8.6
KNN	Caco-2	85.6	8.4
	CYP3A4	87.9	7.8
	hERG	80.5	8.9
	HOB	74.2	14.7
	MN	81.0	15.2
NB	Caco2	82.5	11.5
	CYP34	85.6	10.1
	hERG	744	15
	HOB	56.5	32.4
	MN	72.4	23.8