Skip to main content

Table 3 Overall performance comparison Performance of each feature extraction algorithms averaged across data sets. Balanced accuracy (BACC) reported in increasing order.

From: Feature selection and nearest centroid classification for protein mass spectrometry

  Average BACC (+/-) OC-H4 OC-WCX2a OC-WCX2b PC-H4 PC-IMAC-Cu
PCA 0.684 0.139 0.712 0.682 0.893 0.516 0.619
No FE 0.771 0.044 0.763 0.773 0.834 0.777 0.711
T-test 0.804 0.100 0.747 0.965 0.834 0.709 0.766
NSC(20) 0.808 0.148 0.621 0.944 0.973 0.736 0.764
PCA/LDA 0.808 0.137 0.727 0.899 1.000 0.667 0.748
SBS 0.814 0.070 0.823 0.854 0.903 0.729 0.760
P-test 0.837 0.114 0.763 0.944 0.975 0.728 0.773
KS-test 0.838 0.115 0.702 0.929 0.983 0.784 0.791
SFS 0.863 0.103 0.747 0.949 0.991 0.827 0.798
Boosted 0.883 0.070 0.884 0.914 0.982 0.810 0.826
Boosted FE 0.927 0.057 0.854 0.965 1.000 0.906 0.911