An adaptive optimal ensemble classifier via bagging and rank aggregation with applications to high dimensional data

Table 2 Simulated microarray data

	Accuracy	Sensitivity	Specificity	AUC
linear SVM	0.902200	0.907600	0.896800	0.967464
	(0.00451)	(0.00683)	(0.00679)	(0.00216)
polynomial SVM	0.506200	0.716400	0.296000	0.498772
	(0.00383)	(0.05493)	(0.05477)	(0.00640)
radial SVM	0.773200	0.882000	0.664400	0.833576
	(0.03090)	(0.02851)	(0.04473)	(0.03750)
sigmoid SVM	0.905000	0.910400	0.899600	0.968472
	(0.00432)	(0.00655)	(0.00581)	(0.00210)
greedy	0.671400	0.807200	0.535600	0.702040
	(0.04177)	(0.03811)	(0.05508)	(0.05016)
Ensemble	0.900600	0.902400	0.898800	0.968156
	(0.00366)	(0.00661)	(0.00592)	(0.00213)

Average accuracy, sensitivity, specificity and AUC for 50 datasets from the simulated microarray data with N = 100 and d = 5000. Standard errors are reported in parentheses. A single SVM classifier was used with four different kernel settings.

ISSN: 1471-2105