fig6

Figure 6. Data sampling and model construction for the Chang-Phase dataset. (A) Exploration of Chang-Phase data subsets and algorithm combinations, along with the 10-fold Accuracy of the baseline model; (B) Error performance of six algorithms on the validation and test sets; (C) Confusion matrix for the best generalization model, CBC, across the training, validation, and test sets; (D) ROC curve for the CBC model and corresponding AUC values for different classes. CBC: Gradient boosting classification; ROC: receiver operating characteristic; AUC: area under the curve.