分类树集成算法在县域金融贷款风险分类评估中的应用  

The Application of Bagging Classification Tree Integration Algorithm in the Risk Classification Appraisal of Peasant Loans

作　　者：周启清[1] 李毓[2] 

机构地区：[1]西安交通大学经济与金融学院,西安710049 [2]信阳师范学院经济与管理学院,河南信阳464000

出　　处：《经济问题》2009年第12期94-97,共4页On Economic Problems

摘　　要：分类树方法是一种由计算机实现,基于统计理论的非参数识别技术。首先给出Bagging分类树集成算法用于风险评估的基本思想和实现步骤,并以河南省农户贷款的数据为例,分别采用事先预留检验数据集方法和V折交叉确认法来验证Bagging分类树集成算法的预测能力。实证结果表明,与分类树算法、线性判别分析方法进行比较,Bagging分类树集成算法在贷款风险分类评估中具有较好的预测能力。Classification tree achieved by computers is nonparametric recognition method based on statistical theory. This paper introduces the basic thought and realized steps of the risk appraisal about Bagging classification tree integration algorithm. Taking Henan province＇s peasant loan as a example, it tests the forecast ability by means of hold - out test date set and v - fold crossvalidation method. The ruselt shows that comparing with classification tree integration algorithm or modified linear discriminant analysis, Bagging classification tree integration algorithm has better forecast ability in the risk classification appraisal of loans.

关 键 词：贷款风险评估 BAGGING 分类树集成算法 

分 类 号：F830.33[经济管理—金融学]