数据挖掘分类算法在信用卡风险管理中的应用  

Application of Data Mining in Risk Management of Credit Card

作　　者：覃梅 

机构地区：[1]广东省信息网络有限公司,广州510031

出　　处：《现代计算机》2013年第13期13-16,共4页Modern Computer

摘　　要：对信用卡客户进行潜在风险识别对于银行的资金安全具有重要的意义。利用分类算法对客户进行分类是一种可行的解决方案。利用决策树、朴素贝叶斯、支持向量机等多种成熟的分类算法对信息卡数据集进行了分析,在Model Building Time、Overall Accuracy、TP Rate、FPRate、Kappa statistic和ROC Area六个统计量上对它们的效果进行比较。为银行在信用卡审批时,使用数据挖掘分类算法从大量客户中辨识出风险客户提供实验依据。It＇s very important to identify the underling risks of credit card users for the financial safety of the banks. Using classifying algorithms to disambiguate the users is one feasible solution. Uses several state-of-the-art classifying algorithms, such as Decision Tree, Nalve Bayes and Support Vector Machine, to analyze the credit card dataset, and their performances are compared upon 6 indexes： Model Building Time, Overall Accuracy, TP Rate, FP Rate, Kappa statistic and ROC Area. The experiment results provide references for the banks which using the classification algorithm to find out the risky applicants from the large number of applicants in the credit card approval.

关 键 词：信用卡 风险管理 数据挖掘 分类算法 WEKA 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]