Fig. 4: Evaluation of XGBoost model performance.
From: XGBoost model for the quantitative assessment of stress corrosion cracking
a Performance visualisation of nested CV coupled with BHO after evaluating 3000 hyperparameter sets. Here, the red point highlights the optimal XGBoost hyperparameter set, achieving a minimum RMSE of 1.418 ± 0.12. b XGBoost model performance across validation subsets during k-fold CV, with a MAE range from 0.75 to 0.94. c Overall performance of the XGBoost regression model, demonstrating the close fit between actual and predicted values of α values. Key metrics highlight the model’s predictive proficiency, with a high R2 value of 0.949 and a low MAE value of 0.449, while an RMSE value of 0.663 indicates larger prediction deviations. The maximum error predicted was 1.408. All results and metrics are based on the Box-Cox transformed values of α.
