摘要
数据挖掘技术为商业银行信用风险管理问题提供了新的思路和方法。本文运用三种常用的数据挖掘方法——多元判别分析、聚类分析及贝叶斯网络模型,以商业银行的客户信用风险评级指标数据为样本,对信用风险评估方法进行实证分析,对三种方法的验证结果进行比较。结论表明,在信用风险各项属性指标之间条件相互依赖的情况下,贝叶斯网络模型优于其它两种方法。
Data mining technology provides a new idea and method for the credit risk management of commercial banks. This paper compares three kinds of common data mining methods -- multiple discriminant analysis, cluster analysis and the Bayesian network model, which are applied in credit risk evaluation of commercial banks, using the real data sample of a national commercial bank.The comparison result shows that, under the condition of mutual dependence among the credit risk indicators, Bayesian network model is better than the other two methods.
出处
《网络安全技术与应用》
2013年第3期76-79,共4页
Network Security Technology & Application
关键词
信用风险
多元判别分析
聚类分析
贝叶斯网络
Credit risk
Multiple discriminant analysis
Cluster analysis
Bayesian network
