Figure 2.

Heatmap representing the predictive performance (AUC) of feature selection (rows) and classification (columns) methods in the validation set for: (a) simulated samples with a high signal; (b) simulated samples with a low signal. HC + WLCX = Hierarchical clustering + Wilcoxon; KNN = K-Nearest Neighbor; LSR = Logistic Step-wise Regression; MRMR = Minimum Redundancy Maximum Relevance; PCA (Delta) = Principal Component Analysis clustering with Delta plot stop criterion + Wilcoxon; PCA (Proportion) = Principal Component Analysis clustering with proportion stop criterion + Wilcoxon; PR = Penalized Regression; RF = Random Forest; RELF = Relief; XGBoost = Extreme Gradient Boosting; SVM = Support Vector Machine.