Figure 3.

Box plots and significance map for comparison of Low vs High severity model performance (ROC AUC value) observed in cross validation results of all feature-specific models and combinations. Bottom right panel represents the results of statistical significance test on ROC AUC values performed for each pair of models, where a green star represents rejection of null hypothesis at p < 0.05 (significant performance difference), while a red star indicates the contrary. Wilcoxon signed rank test was performed for each comparison and p-values were corrected through BH correction. Order of the stars (each indicating statistical significance of comparison in a given geography) maps with the geographical regions namely Asia, Europe, North America, Africa and South America.