*4.4. The Influence of the Parameter Setting in CT-XGBoost*

As mentioned in Section 3.2, our proposed CT-XGBoost model has modifications in the form of two algorithm-level methods: the cost-sensitive strategy and the threshold method. The parameter in the cost-sensitive strategy is the penalty ratio, which can assign different misclassification costs to different class samples, and the parameter in the threshold method is the threshold value, which can be used to classify the default probabilities into two classes. Considering that these two parameters can influence the performance of CT-XGBoost, we further analyzed how the credit default performance changes with different parameters and found the best parameter settings.
