基于XGBoost与LR融合模型的信用卡欺诈检测
A fusion model for credit card fraud detection based on XGBoost and LR
张海洋 1陈玉明 1曾念峰 2卢俊文1
作者信息
- 1. 厦门理工学院 计算机与信息工程学院,福建 厦门 361024
- 2. 易成功(厦门)信息科技有限公司,福建 厦门 361024
- 折叠
摘要
随着银行卡业务的不断发展,各种各样的信用卡欺诈方式已经给金融机构带来严重的威胁,使得信用卡欺诈检测成为一个十分紧迫的任务.为解决此问题,提出一种XGBoost与LR融合模型.该模型首先运用XGBoost算法自动进行特征组合和离散化,然后将新构造的特征向量运用在逻辑回归LR模型上,通过XGBoost与LR融合模型进行分类预测.实验结果表明,与经典传统算法相比,提出的XGBoost与LR融合模型具有更好的欺诈检测性能,提高了信用卡欺诈检测的准确率.
Abstract
With the continuous development of bank card business, a variety of credit card frauds pose serious threats to financial institutions. Thus, the early detection of credit card fraud is an urgent task. This paper proposes a fusion model of XGBoost and LR. The model first employs XGBoost algorithm to automatically combine and discretize features, and then applies the newly constructed feature vector to the Logistic Regression model for classification prediction. Our experimental results show, the proposed XGBoost and LR fusion model delivers better fraud detection performance and improves the accuracy of the detection of credit card frauds.
关键词
XGBoost/欺诈检测/逻辑回归/融合模型/信用卡Key words
XGBoost/fraud detection/logistic regression/fusion model/credit card引用本文复制引用
基金项目
国家自然科学基金项目(61976183)
中央引导地方发展科技专项(2022L3029)
出版年
2024