电信客户流失预警规则及其信度测定实证研究——以云南电信为例  

An Empirical Study on the Early Warning Rules of Telecom Customer Churn and Its Reliability Determination——Taking Yunnan's Telecom Industry as an Example

作　　者：顾光同[1,2] 王力宾[2] 费宇[2] 

机构地区：[1]浙江农林大学理学院,杭州311300 [2]云南财经大学统计与数学学院,昆明650221

出　　处：《云南财经大学学报》2010年第6期94-98,共5页Journal of Yunnan University of Finance and Economics

基　　金：云南省教育厅科研基金项目"电信客户流失的统计建模与预测"(09J0011)

摘　　要：以云南省昆明市区电信行业客户数据为样本,应用决策树算法分析流失客户特征,并得出电信客户流失的预警规则,计算出其预测准确率,再在显著的客户信息和预警规则的条件下,测定出所提取的规则发生的概率,即信度。实证研究表明:提取的预警规则准确率和信度都在95%以上,这为电信行业提供了有效的客户挽留决策分析依据。By taking the customer data of telecom industry in Kunming city of Yunnan Province as samples, the authors use decision tree algorithm to analyze the characteristics of loss customers, find out the early warning roles of telecom customer chum, and calculate the predicting accuracy. Besides, based on the significant customer information and the early warning rules, the authors measure the probability that the roles may occur, that is, the reliability. The empirical study shows that the accuracy and the reliability of the early warning roles are above 95%, which provides effective reference to customer retention analysis for telecom industry.

关 键 词：电信客户流失 决策树 预警规则 准确率 信度 

分 类 号：F224.7[经济管理—国民经济]