Comparison of five supervised feature selection algorithms leading to top features and gene signatures from multi-omics data in cancer

Table 5 Average redundancy rate in terms of Pearson correlation coefficient (denoted as \(RR_{avg}^{pc}\)) of different subsets of features selected using various algorithms. Least value of \(RR_{avg}^{pc}\) signifies better choice of the feature selection algorithm

Dataset	Feature selection algorithm
Dataset	mRMR	INMIFS	DFS	SVM-RFE-CBR	VWMRmR
Exp	0.0281	0.0345	0.0145	0.0124^a	0.0219
ExpExon	0.0258	0.0378	0.0177^a	0.0545	0.025
hMethyl27	0.109	0.1235	0.1324	0.0754^a	0.084
Gistic2	0.2953	0.1665	0.3244	0.2437	0.1277^a
Pathway activity data	0.042	0.0291	0.0467	0.0684	0.025^a

^aBold font justifies least \(RR_{avg}^{pc}\) among all algorithms (i.e., best performance) for each data (row)

ISSN: 1471-2105