Credit Card Fraud Detection Method Based on KRSMOTE+ENN and XGBoost Algorithm
Lingfei Ma · 2024
With the rapid development of credit card business of financial institutions, credit card fraud has become a serious problem. For the problem of unbalanced distribution of credit card data, this paper adopts a variety of sampling methods to balance the unbalanced data, and then inputs the balanced dataset into a variety of classification algorithms for experimental comparison, and finally proposes a credit card fraud detection model based on the hybrid sampling of KRSMOTE + ENN and the XGBoost algorithm. The detection method is verified by five evaluation indexes that not only improves the differentiation of unbalanced data of credit card fraud behavior, but also improves the accuracy and feasibility of credit card fraud detection.