A comprehensive comparison of random forests and support vector machines for microarray-based cancer classification

Table 2 Comparison of classification performance of SVMs and RFs with gene selection.

*Task & dataset*	*Classification performance metric*	*Classification performance*		*Nominally superior method*	*P-value*
		*SVM*	RF
Dx-Alon	AUC	0.938	0.917	SVM	0.626
Dx-Ramaswamy2	AUC	0.821	0.781	SVM	0.624
Dx-Shipp	AUC	0.992	0.975	SVM	0.502
Dx-Singh	AUC	0.964	0.972	RF	0.812
Px-Beer	AUC	0.798	0.648	SVM	0.016
Px-Bhattacharjee	AUC	0.519	0.561	RF	0.550
Px-Iizuka	AUC	0.713	0.763	RF	0.750
Px-Pomeroy	AUC	0.692	0.629	SVM	0.506
Px-Rosenwald	AUC	0.689	0.631	SVM	0.128
Px-Veer	AUC	0.758	0.754	SVM	0.954
Px-Yeoh	AUC	0.777	0.716	SVM	0.082
Dx-Alizadeh	RCI	1.000	1.000	-	1
Dx-Armstrong	RCI	0.944	0.911	SVM	0.624
Dx-Bhattacharjee	RCI	0.895	0.817	SVM	0.125
Dx-Golub	RCI	0.953	0.934	SVM	1
Dx-Khan	RCI	1.000	1.000	-	1
Dx-Nutt	RCI	0.812	0.733	SVM	0.220
Dx-Pomeroy	RCI	0.823	0.688	SVM	0.079
Dx-Ramaswamy	RCI	0.911	0.880	SVM	0.066
Dx-Staunton	RCI	0.876	0.856	SVM	0.626
Dx-Su	RCI	0.958	0.922	SVM	0.078
Px-Veer2	RCI	0.451	0.371	SVM	0.262

The performance is estimated using area under ROC curve (AUC) for binary classification tasks and relative classifier information (RCI) for multicategory tasks. See subsection "Statistical comparison among classifiers" for the description of statistical test employed to compute reported p-values. P-values shown with boldface denote statistically significant differences between classification methods at the 0.05 α level.

ISSN: 1471-2105