基于SMOTEENN-XGBoost的信用卡风险客户预测  

Prediction of Customers with Credit Card Risk Based on SMOTEENN-XGBoost

作　　者：田园 郭红烈 吉倩 TIAN Yuan;GUO Hongie;JI Qian(Department of Economic and Trade,Qujing Vocational and Technical College,Qujing 655000,China;School of Informa-tion Engineering and Automation,Kunming University of Science and Technology,Kunming 650500,China)

机构地区：[1]曲靖职业技术学院经济贸易系,云南曲靖655000 [2]昆明理工大学信息工程与自动化学院,云南昆明650500

出　　处：《软件导刊》2024年第7期138-143,共6页Software Guide

基　　金：云南省基础研究计划面上项目(202201AT070189)。

摘　　要：为了实现信用卡的风险管控,降低因信用卡违约造成的经济损失,构建有效的信用卡风险预测模型尤为重要。针对信用卡数据分布不均衡的问题,使用ENN算法对经典SMOTE算法进行改进,构建了基于SMOTEENN-XGBoost的信用卡风险预测模型。实验表明,该模型的预测准确率能达到91.8%、AUPRC值为0.903,显著优于SVC、GBDT、AdaBoost等经典模型,对于信用不良信用卡用户的预测、帮助银行准确甄别客户信用风险具有重要价值。To achieve risk management for credit cards and reduce economic losses caused by credit card defaults,it is particularly important to develop an effective credit card risk prediction model.In response to the issue of imbalanced credit card data distribution,the ENN algo-rithm was used to improve the classical SMOTE algorithm,resulting in the construction of a credit card risk prediction model based on SMO-TEENN-XGBoost.Empirical evidence reveals that this model achieves a prediction accuracy of 91.8%and an AUPRC value of 0.903,which is significantly better than classical models such as SVC,GBDT,and AdaBoost.It holds significant value in predicting high-risk credit card users and aiding banks in accurately identifying customer credit risks.

关 键 词：信用卡风险预测 数据平衡 SMOTEENN XGBoost 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]