A Comparative Study on Credit Card Default Risk Predictive Model

Abstract

The risk analysis of credit card defaulters is a critical procedure in the banking sector to classify the card applicants. Banks perform credit score check to make decisions on applications and to set credit limit accordingly. With the increase in the amount of data and advances in data analytics, the approval process can now be automated for quicker processing of applications. This study aims to provide solutions to improve the risk management strategy among financial institutions using predictive analytics. A real-world dataset obtained from a bank in Taiwan were used to perform the analysis in this project. Four data mining algorithms including Decision Tree, Logistic Regression, Random Forest and Neural Network were constructed with the cleaned dataset. Results revealed that Neural Network is the best performing model with an 82% predictive accuracy of credit card defaulters.