A new regularized least squares support vector regression for gene selection

BMC Bioinformatics

Table 7 Testing accuracies under different procedures for SRBCT data

Procedures	Classifier	No. of genes	Accuracy
A: Proposed selection and criterion
RLS-SVR
δ_j≥ 1/q	+KFDA	2	0.6
	+SVM	2	0.55
∑ δ_j≥ 80%	+KFDA	8	0.95
	+SVM	8	0.95
q genes	+KFDA	10	1
	+SVM	10	1
B: Other selection procedures
EB	+KFDA	14	1
	+SVM	14	1
SGS1	+KFDA	10	0.8
	+SVM	10	0.7
SGS2	+KFDA	10	0.85
	+SVM	10	0.85
C: Selection and classification together
	EB	14	1

ISSN: 1471-2105