摘 要：近些年信用卡的违约情况呈现逐年上升的趋势，使商业银行面临严重的经营风险，商业银行若想在信用卡业务中获得利润，必须控制信用卡的违约率。关于信用卡违约的研究主要围绕信用评级展开，鉴于传统单一分类器预测模型拟合不足或过拟合的缺陷，提出改进后的随机森林预测模型，并在实证分析中与 KNN、逻辑回归、决策树和 GBDT 相比较。模型提高了信用卡违约识别率，降低了违约风险，对提高商业银行的风险管控能力具有积极意义。
关键词：信用卡违约；逻辑回归；GBDT；ROC 曲线；随机森林
中图分类号：TP391 文献标识码：A 文章编号：2096-4706（2020）03-0001-05
Research on Credit Card Default Prediction Based on Random Forest
GUO Jianshan，QIAN Junhao
（School of Internet of Things Engineering，Jiangnan University，Wuxi 214122，China）
Abstract：In recent years，the default situation of credit cards has been increasing year by year，which makes commercial Banks face serious operational risks. If commercial banks want to gain profits from credit card business，they must control the default rate of credit cards. The research on credit card default mainly focuses on credit rating. In view of the deficiency or over fitting of the traditional single classifier prediction model，an improved stochastic forest prediction model is proposed and compared with KNN，logistic regression，decision tree and GBDT in the empirical analysis. The model improves the credit card default recognition rate and reduces the default risk，which is of positive significance to improve the risk control ability of commercial Banks.
Keywords：credit card default；logistic regression；GBDT；ROC curve；random forest
参考文献：
[1] BRAUSE R，LANGSDORF T，HEPP M. Neural data mining for credit card fraud detection [C]// Tools with Artificial Intelligence，1999. Proceedings. 11th IEEE International Conference on，1999：103-106.
[2] JERAGH M，ALSULAIMI M. Combining Auto Encoders and One Class Support Vectors Machine for Fraudulant Credit Card Transactions Detection [C]// 2018 Second World Conference on Smart Trends in Systems，Security and Sustainability （WorldS4），London，England：2018：178-184.
[3] 邹权 . 基于数据挖掘的信用卡申请者信用评分模型研究 [D]. 成都：西南财经大学，2004.
[4] 张双全 . 基于改进智能算法的信用卡客户违约预测研究 [D]. 长春：长春工业大学，2018.
[5] 余以胜 . 基于随机森林的用户行为识别模型研究 [J]. 电脑知识与技术，2017，13（7）：156-157.
[6] KUMAR M. S，SOUNDARYA V，KAVITHA S，et al. Credit Card Fraud Detection Using Random Forest Algorithm [C]// 2019 3rd International Conference on Computing and Communications Technologies （ICCCT），Chennai，India：2019：149-153.
[7] XIE Y，LIU G，CAO R，et al. A Feature Extraction Method for Credit Card Fraud Detection [C]// the 2ndIEEE International Conference on Intelligent Autonomous Systems（ICoIAS’2019），Singapore，2019：70-75.
[8] MITTAL S，TYAGI S. Performance Evaluation of Machine Learning Algorithms for Credit Card Fraud Detection [C]// 2019 9th International Conference on Cloud Computing，Data Science & Engineering （Confluence），Noida，India：2019：320-324.
[9] MACHADO M. R，KARRAY S，SOUSA I. T. LightGBM：an Effective Decision Tree Gradient Boosting Method to Predict Customer Loyalty in the Finance Industry [C]// 2019 14th International Conference on Computer Science & Education （ICCSE），Toronto，ON，Canada：2019：1111-1116.
[10] GOY G，GEZER C，GUNGOR V. C. Credit Card Fraud Detection with Machine Learning Methods [C]// 2019 4th International Conference on Computer Science and Engineering （UBMK），Samsun，Turkey：2019：350-354.
作者简介：郭建山（1995.11-），男，汉族，福建莆田人，硕士研究生，研究方向：数据挖掘。