Fraud Prediction of Credit Card Customers Based on Xgboost Model and Multi-Layer Perception Model
Zhu Xueping, Qingnian Li, Huang Ying, Lei Huang, Deng Pengying · 2022 IEEE International Conference on Advances in Electrical Engineering and Computer Applications (AEECA) · 2022
With the boom in mobile web applications and the electronic payment industry, the use of credit cards for fraudulent purposes is increasing. How to handle credit card fraud in a timely manner is now the pressing challenge for financial institutions and operators to overcome. To address this problem, this paper conducts experiments using the credit card public dataset available on Kaggle. For the uneven distribution of data samples in the dataset, this paper uses the down sampling method, followed by feature filtering by ranking the importance of the features in the dataset, and finally using XGoost and the multilayer perceptron to analyse and predict credit card fraud in the dataset. This study tests the performance of the two models using accuracy, recall, and f1-score in order to fairly assess how well they perform on this issue. Empirical evidence shows that the XGBoost model outperforms the multilayer perceptron model in the credit card fraud problem.