Fig. 5
From: Random forest versus logistic regression: a large-scale benchmark experiment
Subgroup analyses. Top: for each of the four selected meta-features n, p, p/n and Cmax, boxplots of Δacc for different thresholds as criteria for dataset selection. Bottom: distribution of the four meta-features (log scale), where the chosen thresholds are displayed as vertical lines. Note that outliers are not shown here for a more convenient visualization. For a corresponding figure including the outliers as well as the results for auc and brier, see Additional file 1