自我网络特征对电信客户流失的影响

发布时间：2018-05-11 00:20

本文选题：社交网络 + 度　；参考：《管理科学》2017年05期

【摘要】：近年来,随着移动通信行业的蓬勃发展,市场饱和度越来越高,企业获取新用户的成本也越来越大。随着中国三大运营商竞争的加剧,产品和服务的同质化程度也越来越高,这使企业在老客户的保留上变得异常困难,客户流失率也在逐年上升,如何识别高风险流失客户并有效防止客户流失已经成为该行业管理者普遍关心的问题之一。着眼于客户流失影响因素研究,运用社交网络分析方法,通过构造与网络结构相关的变量进行影响因素的探讨,运用逻辑回归方法构建客户流失预警模型。从社交网络的视角出发,利用客户的通话详单数据建立客户之间的通信网络,在自我网络的相关理论框架下,构建个体的度、联系强度、个体的信息熵3个自我网络特征变量。运用中国某移动运营商公司的月度客户数据(包括基础通信数据和通话详单数据),通过逻辑回归构建基于社交网络变量的客户流失预警模型。研究结果表明,个体的度、联系的强度和个体的信息熵都对预测客户流失有显著效果。具体的,个体的度越大,联系强度越强,个体的信息熵越大,客户越不容易流失。外样本AUC值平均可以达到0.75以上,模型具有良好的预测精度。研究结果对企业实践具有非常重要的意义,合作企业应用客户流失预警模型进行高风险流失客户的识别,预测精度可以达到70%,达到了企业的实践预期。客户流失预警模型可以帮助企业提前识别高风险流失客户,极大地降低企业维系客户的成本。建议企业管理者在未来更加关注与客户社交网络有关的变量,从网络结构的视角理解消费者行为,更好地进行客户关系管理。
[Abstract]:In recent years, with the rapid development of mobile communication industry, the market saturation is becoming higher and higher. With the intensification of competition among the three major operators in China, the degree of homogenization of products and services is becoming more and more high, which makes it extremely difficult for enterprises to retain their old customers, and the rate of customer turnover is also increasing year by year. How to identify high risk customers and effectively prevent them has become one of the problems that managers in this industry are concerned about. Based on the research of influencing factors of customer churn, this paper uses the method of social network analysis to discuss the influencing factors by constructing variables related to the network structure, and constructs the early warning model of customer churn by logical regression method. From the point of view of social network, the communication network between customers is established by using customer's call detail sheet data. Under the relevant theoretical framework of self-network, three self-network characteristic variables, degree of individual, contact intensity and individual information entropy, are constructed. Based on the monthly customer data (including basic communication data and call detail data) of a mobile operator in China, a customer churn warning model based on social network variables is constructed by logical regression. The results show that the degree of individual, the intensity of connection and the information entropy of individual have significant effect on predicting customer turnover. Specifically, the greater the degree of the individual, the stronger the connection intensity, the greater the information entropy of the individual, the more difficult the customer is to lose. The average AUC value of the external sample can reach more than 0.75, and the model has good prediction accuracy. The research results are of great significance to the practice of enterprises. The prediction accuracy can reach 70 points by applying the early warning model of customer churn to identify the high risk customers, which can reach the expectation of enterprise practice. Customer churn warning model can help enterprises identify high risk customers in advance and greatly reduce the cost of maintaining customers. It is suggested that enterprise managers pay more attention to the variables related to customer social networks in the future, understand consumer behavior from the perspective of network structure, and better carry out customer relationship management.
【作者单位】：中国人民大学统计学院;上海科技大学创业与管理学院;北京大学光华管理学院;
【基金】：中国人民大学科学研究基金重大规划项目《互联网统计学研究》资助~~
【分类号】：F274;F626

【相似文献】