基于决策树的证券客户流失模型  

THE STOCK CUSTOMERS CHURN MODEL BASED ON DECISION TREE

作　　者：杜修平[1] 王中 

机构地区：[1]天津大学现代教育技术研究所,天津300072 [2]中信证券公司,天津300072

出　　处：《计算机应用与软件》2009年第9期230-233,共4页Computer Applications and Software

摘　　要：描述了证券业客户流失分析的重要性,客户流失的定义,提出了影响客户流失的各种特征因素。然后根据CRM中的RFM模型,加入客户收益率指标,提出了证券行业客户流失分析的RFM-ROI模型。用决策树方法构建了客户流失分析模型,并提出了解决决策树剪枝问题的停止阈值方法。结果表明该模型能达到80.7%的准确率,具有较强的实用性。The definition of customers churn and the importance of analyzing stock customers churn were described in this paper. And several characteristic factors affecting customers churn were proposed. According to the RFM model in Customer Relationship Management （CRM） theory, and customer Return on Investment, the RFM-ROI model was proposed. In the paper, a customer churn analyzing model was constructed with decision tree, and an optimal stopping threshold method was put forward, which could act as a tree pruning strategy. The resuits showed the model has 80.7% accuracy, and it has real use.

关 键 词：证券 RFM—ROI 模型决策树 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] F840.4[自动化与计算机技术—控制科学与工程]