Research on Credit Card Risk Management Based on Decision Tree

Li Yangzhi

Abstract:Since the introduction of credit cards in China, its development speed is increasing.With the development of China's economy and changes in national consumption habits, credit card business will become an indispensable part of people's lives.Credit risk management is particularly important in order to make credit cards better for economic development and people's lives.Because credit risk is the main risk of credit card business. Therefore, the issuer should strengthen the control of credit risk management, the establishment of a reasonable and effective, scientific and sound credit evaluation system. In general, credit ratings will divide customers into“good”or“bad”customers.According to the customer's personal information and historical data, through mathematical methods for modeling, and then to predict the credit card customer default risk.In this respect, the decision tree algorithm is accurate and intuitive.This paper uses decision tree as a method to establish credit evaluation model.This paper first summarizes the literatures involved, combines the credit card risk management theory and the theoretical framework of decision tree, and analyzes the characteristics of credit risk in decision tree and credit card business.The decision tree C4.5, C5.0, rpart, xgboost and combined forecasting model are used to model and compare with the support vector machine.The personal credit evaluation model constructed by the decision tree method has many advantages such as high precision, strong handling and wide use in practice.

Key Words:Credit Card; Decision Tree; Risk Management