Skip to main content

Table 2 Detailed performance statistics for prostate cancer data sets Bold columns represent the mean of the respective performance measure, while columns labeled as (std) correspond to the standard deviation across the three cross-validation folds.

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

PC-H4 Corr Corr(std) BACC 3ACC(std Spec Spec(std) Sens Sens(std) PPV PPV(std)
No FE 0.732 0.05 0.777 0.06 0.698 0.05 0.855 0.09 0.439 0.06
PCA 0.530 0.20 0.516 0.18 0.540 0.24 0.493 0.21 0.248 0.11
PCA/LDA 0.692 0.15 0.667 0.14 0.710 0.22 0.623 0.33 0.431 0.17
SFS 0.885 0.05 0.827 0.17 0.929 0.03 0.725 0.36 0.728 0.03
SBS 0.773 0.03 0.729 0.09 0.806 0.11 0.652 0.27 0.498 0.07
P-test 0.813 0.02 0.728 0.11 0.877 0.08 0.580 0.28 0.572 0.07
T-test 0.816 0.04 0.709 0.14 0.897 0.05 0.522 0.31 0.575 0.07
KS-test 0.826 0.04 0.784 0.14 0.857 0.08 0.710 0.35 0.579 0.05
NSC(20) 0.791 0.04 0.736 0.10 0.833 0.12 0.638 0.31 0.529 0.07
Boosted 0.850 0.06 0.810 0.11 0.881 0.04 0.739 0.22 0.627 0.10
Boosted FE 0.960 0.01 0.906 0.03 1.000 0.00 0.812 0.07 1.000 0.00
PC-IMAC-Cu Corr Corr(std) BACC 3ACC(std Spec Spec(std) Sens Sens(std) PPV PPV(std)
No FE 0.709 0.13 0.711 0.13 0.767 0.14 0.655 0.12 0.750 0.15
PCA 0.618 0.07 0.619 0.07 0.654 0.21 0.583 0.20 0.652 0.08
PCA/LDA 0.746 0.03 0.748 0.03 0.818 0.07 0.679 0.04 0.800 0.06
SFS 0.795 0.03 0.798 0.03 0.912 0.14 0.685 0.07 0.914 0.13
SBS 0.758 0.15 0.760 0.15 0.818 0.15 0.702 0.15 0.802 0.17
P-test 0.771 0.06 0.773 0.06 0.843 0.15 0.702 0.05 0.840 0.12
T-test 0.765 0.06 0.766 0.06 0.805 0.10 0.726 0.09 0.803 0.08
KS-test 0.789 0.03 0.791 0.03 0.862 0.09 0.720 0.05 0.854 0.07
NSC(20) 0.761 0.09 0.764 0.09 0.868 0.12 0.661 0.17 0.849 0.10
Boosted 0.823 0.05 0.826 0.05 0.950 0.09 0.702 0.11 0.949 0.09
Boosted FE 0.908 0.02 0.911 0.02 1.000 0.00 0.821 0.03 1.000 0.00