Fig. 2From: Variable selection for binary classification using error rate p-values applied to metabolomics dataSimulation comparison of the different error rate test statistics. Figures a (weight set 1) and b (weight set 2) depict the average p-values associated with \( {\widehat{er}}_{up}^{*} \) (red lines), \( {\widehat{er}}_{min}^{*} \) (blue lines) and the MW test statistic (black lines). Figures c (weight set 1) and d (weight set 2) depict the proportions of repetitions in which the p-values of \( {\widehat{er}}_{min}^{*} \) (blue lines) and MW (black lines) were below the p-values of \( {\widehat{er}}_{up}^{*} \). The dotted red line represents the 50Â % cut-off. The dashed blue lines represent points of reference as discussed in the textBack to article page