贝叶斯网络在预测银行信贷风险中的应用  

Application of Bayesian Network in Forecasting Bank Loan Defaulters

作　　者：周森鑫[1] 李超[1] 吴德成[1] 

机构地区：[1]安徽财经大学管理科学与工程学院,安徽蚌埠233000

出　　处：《鸡西大学学报（综合版）》2014年第10期42-43,63,共3页JOurnal of Jixi University:comprehensive Edition

基　　金：安徽省高校自然科学基金重大项目(ZD200905);安徽省高校教学研究项目(20100473)

摘　　要：对银行来说,分析量化操作风险,对管理者进行相关决策十分重要。贝叶斯网络能够很好地表达不确定因素之间的因果关系,并用于推理决策,是提高操作风险管理水平的有效途径。通过构建贝叶斯网络模型,将可能导致贷款者拖欠还款的各种影响因素引入到具有因果关联的网络结构中,计算出各类指标对拖欠的影响程度。并且通过分析节点,直观地显示了每个模型的预测效果,为银行预测潜在贷款拖欠者,并进行相关风险管理提供了理论依据。The quantification of operational risk is very important for bank managers in making decision. Bayesian network can not only express the causal relationship between uncertainty factors well,but some reasoning,which is an efficient way to improve operational risk management.,We introduce various factors affected the network structure with causal correlation By building a Bayesian network,and calculate the effect of every factor for loan defaulters. After evaluating every node,we can understand the predictive effect of each model Intuitively,which is helpful for banks to forecast loan defaulters and provide a theoretical basis for the risk control.

关 键 词：贝叶斯网络 商业银行 贷款拖欠者 风险管理 

分 类 号：TP31[自动化与计算机技术—计算机软件与理论]