Figure 3.

Boxplots of the false positive rates (over the 50 repeated cross-validation testing sets) for each feature selection method for the four best-performing classifiers.