Prediction of Consumer Credit Risk Via Machine Learning Algorithm

This thesis includes research on the development of a credit score model for assessing creditworthiness. The work focused primarily on the issue of imbalanced class credit datasets, which can have a negative impact on the performance of the credit score model. In an effort to address the issue, this study attempted to create an optimized ensemble model that combines the resampling technique, ensemble classifier, and dynamic threshold. In addition, the technique for interpreting the prediction output was explored.