Figure S1.
The ROC curve for XGBoost with fivefold cross-validation and a hold-out test. The x-axis represents the false positive rate (FRP) ranging from 0.00 to 1.00, and the y-axis represents the true positive rate (TPR) also ranging from 0.00 to 1.00. The area under the curve (AUC) is 0.551 with a standard deviation of 0.145. The ROC curve for LightGBM with similar axes and an AUC of 0.550 with a standard deviation of 0.100. The ROC curve for CatBoost with an AUC of 0.683 and a standard deviation of 0.250. The ROC curves of the hold-out test across the three classifiers, with XGBoost having an AUC of 0.5, LightGBM an AUC of 0.83, and CatBoost an AUC of 0.92.
Predictive performance of gradient boosting classifiers. Receiver operating characteristic (ROC) curves are shown for XGBoost, LightGBM, and CatBoost models. (A–C) Panels depict ROC curves from fivefold cross-validation (colored lines) along with the independent holdout test (dashed line). (D) Panel summarizes the ROC curves of the holdout test across the three classifiers. FRP, false positive rate; TPR, true positive rate.