Prediction of donor splice sites using random forest with a new sequence encoding approach

Table 3 P-values of Mann Whitney U statistic for testing the significant difference between RF-SVM, RF-ANN and SVM-ANN at 5 % level of significance for all the performance measures under both balanced and imbalanced training datasets

$D	EP	MLA	TPR	TNR	F (α = 1)	F (β = 2)	G-mean	WA	MCC
Balanced	P-1	RF-SVM	0.02008	0.42473	0.32557	0.04117	0.40550	0.38378	0.32557
		RF-ANN	0.00356	0.73286	0.01854	0.00520	0.02323	0.02569	0.01854
		SVM-ANN	0.01696	0.30585	0.07526	0.03546	0.07526	0.09605	0.10512
	P-2	RF-SVM	0.00018	0.02564	0.00001	0.00001	0.00001	0.00001	0.00001
		RF-ANN	0.00018	0.00018	0.00001	0.00001	0.00001	0.00001	0.00001
		SVM-ANN	0.05869	0.54505	0.16549	0.06301	0.14314	0.14017	0.24745
	P-3	RF-SVM	0.00018	0.00066	0.00001	0.00001	0.00001	0.00001	0.00001
		RF-ANN	0.00018	0.00129	0.00001	0.00001	0.00001	0.00001	0.00001
		SVM-ANN	0.93961	0.16150	0.10512	0.68421	0.07526	0.06954	0.07526
Imbalanced	P-1	RF-SVM	0.00018	0.00018	0.00001	0.00001	0.00001	0.00001	0.00001
		RF-ANN	0.00017	0.00017	0.00001	0.00001	0.00001	0.00001	0.00001
		SVM-ANN	0.00048	0.46778	0.00008	0.00008	0.00008	0.00008	0.00021
	P-2	RF-SVM	0.00018	0.00017	0.00001	0.00001	0.00001	0.00001	0.00001
		RF-ANN	0.00018	0.00018	0.00001	0.00001	0.00001	0.00001	0.00001
		SVM-ANN	0.64854	0.05130	0.39305	0.52885	0.48125	0.32557	0.05243
	P-3	RF-SVM	0.00018	0.00018	0.00001	0.00001	0.00001	0.00001	0.00001
		RF-ANN	0.00018	0.00018	0.00001	0.00001	0.00001	0.00001	0.00001
		SVM-ANN	0.49483	0.05210	0.63053	0.57874	0.57874	0.73936	0.91180

$D type of dataset (balanced or imbalanced), EP encoding procedures (P-1, P-2, P-3), MLA machine learning approaches

ISSN: 1756-0381