A Credit Card Fraud Detection Method Based on Mahalanobis Distance Hybrid Sampling and Random Forest Algorithm

Abstract

Currently, the continuous expansion of the credit card business and the increasingly fierce competition have made all kinds of fraud risks in the overall credit card business the biggest threat. How to detect credit card fraud accurately and effectively through machine learning algorithms has become a research hotspot and challenge in this field. This project addresses the problems of existing credit card fraud detection methods and, based on the domain knowledge of credit card fraud and machine learning theory, proposes a method based on Mahalanobis distance SMOTE-ENN hybrid sampling and Random Forest for credit card fraud detection. First, fraud detection experiments were conducted by selecting credit card fraud datasets published on the Kaggle platform. Then, to further confirm the effectiveness of the method, the method was used for the credit card customer default dataset in Taiwan for further experimental proof. Finally, the experimental results in this paper were compared with the best experimental results in other articles, reflecting that the method based on Mahalanobis distance SMOTE-ENN hybrid sampling and Random Forest can be more effective for credit card fraud detection compared with other methods. The proposed method demonstrates a strong application potential and practical effect in credit card fraud detection, which not only improves the accuracy and efficiency of fraud detection but also provides a wide range of insights and methodological references for the solution of similar problems.

Keywords

• Credit card fraud
• data balancing processing
• Mahalanobis distance
• SMOTE-ENN hybrid sampling
• random forest algorithm