A Credit Card Default Prediction Method Based on CatBoost

Abstract

This paper presents a study on the prediction of credit card user default using the CatBoost model. The dataset used in this study is a credit card dataset from a financial institution. The dataset contains information about the credit card users such as their age, gender, credit limit, and payment history. The CatBoost model was used to predict the probability of default for each user. The results showed that the CatBoost model was able to accurately predict the probability of default for credit card users. And in the experiment, I found that the prediction effect of CatBoost model is better than that of XGBoost, Lasso, and LightGBM. The results of this study can be used to help financial institutions better manage their credit card portfolios and reduce the risk of default.