Fig 2. Results of the initial model screen.
Plots on the left show the AUROC of the best hyperparameter set of each model on the cross-validation folds. Plots on the right show the AUROC of k randomly generated parameter combinations as measured on the test sets, where k = 10×(number of hyperparameters). Boxes denote data within 25th and 75th percentiles and the middle line represents the median. Whiskers extend from the box to 1.5× the interquartile range.
