Table 4 The results of feature selection on the discovery dataset performed by seven different RCPH models
From: Detecting prognostic biomarkers of breast cancer by regularized Cox proportional hazards models
Methods | Training dataset (\(70 \%\)) | Testing dataset (\(30 \%\)) | |
|---|---|---|---|
# of features | C-index ± Std. Dev | P-value ± Std. Dev | |
Ridge-RCPH | 1142 | \(1.000 \pm 0.000\) | \(0.013 \pm 0.001\) |
Lasso-RCPH | 47 | \(0.726 \pm 0.022\) | \(0.005 \pm 0.007\) |
Enet-RCPH | 66 | \(0.798 \pm 0.044\) | \(0.003 \pm 0.002\) |
\(L_{1/2}\)-RCPH | 4 | \(0.629 \pm 0.000\) | \(0.041 \pm 0.000\) |
\(L_{0}\)-RCPH | 17 | \(0.794 \pm 0.000\) | \(0.000 \pm 0.000\) |
SCAD-RCPH | 42 | \(0.731 \pm 0.046\) | \(0.032 \pm 0.018\) |
MCP-RCPH | 22 | \(0.639 \pm 0.019\) | \(0.062 \pm 0.044\) |