Stable feature selection and classification algorithms for multiclass microarray data

Student, Sebastian; Fujarewicz, Krzysztof

doi:10.1186/1745-6150-7-33

Biology Direct

Table 1 The bootstrap based classification accuracies, stability index and number of reselected genes in all bootstrap samples of the SVM-classifier and the LDA-classifier based on all tested gene selection methods, on the LUNG dataset

From: Stable feature selection and classification algorithms for multiclass microarray data

method			stability index		Reselected genes	Classification method				Best result
						SVM OvO	SVM OvR	MSVM	LDA
			s ₁	s ₂		acc	acc	acc	acc	acc	accL	accH
SIMPLS	OvO	1 comp.	0.86	0.69	84	0.954	0.946	0.951	0.956	0.956	0.930	0.978
	OvR	1 comp.	0.86	0.74	78	0.950	0.945	0.947	0.965	0.965	0.937	0.985
	MCLASS	1 comp.	0.88	0.79	70	0.917	0.882	0.892	0.895	0.917	0.868	0.956
	OvO	5 comp.	0.60	0.51	186	0.950	0.928	0.944	0.966	0.966	0.933	0.986
	OvR	5 comp.	0.71	0.61	170	0.958	0.946	0.953	0.963	0.963	0.940	0.986
	MCLASS	5 comp.	0.75	0.58	103	0.953	0.945	0.949	0.917	0.953	0.920	0.979
	OvO	10 comp.	0.58	0.44	214	0.949	0.928	0.941	0.946	0.949	0.913	0.979
	OvR	10 comp.	0.63	0.49	228	0.955	0.944	0.950	0.961	0.961	0.931	0.983
	MCLASS	10 comp.	0.72	0.48	128	0.955	0.949	0.950	0.943	0.955	0.919	0.985
NIPALS	OvO	1 comp.	0.86	0.69	84	0.954	0.946	0.951	0.956	0.956	0.930	0.978
	OvR	1 comp.	0.86	0.74	78	0.950	0.945	0.947	0.965	0.965	0.937	0.985
	MCLASS	1 comp.	0.88	0.79	70	0.917	0.882	0.892	0.895	0.917	0.868	0.956
	OvO	5 comp.	0.68	0.60	151	0.949	0.932	0.943	0.956	0.956	0.924	0.980
	OvR	5 comp.	0.76	0.62	126	0.954	0.946	0.949	0.954	0.954	0.923	0.980
	MCLASS	5 comp.	0.74	0.62	111	0.949	0.940	0.944	0.930	0.949	0.907	0.979
	OvO	10 comp.	0.66	0.46	143	0.946	0.920	0.936	0.927	0.946	0.907	0.978
	OvR	10 comp.	0.71	0.51	134	0.950	0.938	0.944	0.936	0.950	0.910	0.979
	MCLASS	10 comp.	0.73	0.55	121	0.951	0.944	0.947	0.911	0.951	0.912	0.979
RFE	OvO		0.44	0.35	487	0.962	0.953	0.961	0.977	0.977	0.947	0.999
	OvR		0.30	0.20	808	0.965	0.961	0.964	0.968	0.968	0.938	0.992
T-TEST	OvO		0.09	0.07	333	0.956	0.939	0.951	0.951	0.956	0.923	0.985
	OvR		0.47	0.40	624	0.939	0.925	0.934	0.850	0.939	0.896	0.977
GS	OvO		0.52	0.42	453	0.962	0.944	0.957	0.973	0.973	0.946	0.994
	OvR		0.62	0.49	305	0.964	0.953	0.961	0.971	0.971	0.943	0.994
	MCLASS		0.65	0.53	243	0.935	0.900	0.921	0.943	0.943	0.883	0.974

^aThe number of selected genes is set to 30 In last 3 columns is the best accuracy rate together with their bootstrap based standard deviations (accL,accH). The number of reselected genes is the sum of non zero bootstrap-based feature ranked genes (BBFR).

Back to article page

ISSN: 1745-6150

Contact us

General enquiries: journalsubmissions@springernature.com