NMFBFS: A NMF-Based Feature Selection Method in Identifying Pivotal Clinical Symptoms of Hepatocellular Carcinoma

Table 8

Classification accuracy of inferred optimal feature subset via NMFBFS, ReliefF, mRMR, and Elastic Net on the testing set.

Methods	Feature subset	Dimension	Classification accuracy in LSSVM (%)
NMFBFS		39	79.65 ± 6.48
ReliefF	FSRF20	20	50.71 ± 1.22
	FSRF40	40	76.43 ± 8.27
mRMR	FSMR20	20	63.79 ± 1.22
	FSMR40	40	77.14 ± 9.18
Elastic Net	FSEN20	20	67.57 ± 4.09
	FSEN40	40	78.38 ± 9.62