BMC Bioinformatics

Table 2 The leave-one-out and 5-fold cross validation classification accuracies of the SVM-classifier and the KNN-classifier based on four gene selection methods, GS1, GS2, Cho's, and F-test, on the CAR dataset.

From: A stable gene selection in microarray data analysis

CAR 5-Fold	KNN				SVMs
	30	60	100	Best Accuracy/# Genes	30	60	100	Best Accuracy/# Genes
GS2	0.578 ± 0.118	0.810 ± 0.084	0.865 ± 0.059	0.865/100	0.528 ± 0.116	0.812 ± 0.080	0.870 ± 0.053	0.870/100
GS1	0.634 ± 0.136	0.831 ± 0.079	0.874 ± 0.058	0.874/100	0.600 ± 0.140	0.824 ± 0.076	0.885 ± 0.050	0.885/100
Cho's	0.471 ± 0.091	0.676 ± 0.083	0.797 ± 0.070	0.797/100	0.437 ± 0.089	0.651 ± 0.085	0.821 ± 0.066	0.821/100
F-test	0.681 ± 0.091	0.788 ± 0.071	0.851 ± 0.065	0.851/100	0.649 ± 0.093	0.802 ± 0.071	0.868 ± 0.056	0.868/100
CAR LOO	KNN				SVMs
	30	60	100	Best Accuracy/#Genes	30	60	100	Best Accuracy/# Genes
GS2	0.621	0.828	0.885	0.885/99	0.557	0.822	0.868	0.874/71
GS1	0.718	0.822	0.868	0.879/97	0.695	0.828	0.902	0.902/100
Cho's	0.448	0.661	0.787	0.805/88	0.466	0.661	0.851	0.879/97
F-test	0.707	0.776	0.856	0.862/85	0.626	0.793	0.874	0.885/97

Back to article page

ISSN: 1471-2105

Contact us

General enquiries: journalsubmissions@springernature.com