基于数据挖掘技术的移动存量维系系统设计与实现

发布时间：2018-05-30 20:31

本文选题：存量维系 + 决策树　；参考：《东南大学》2015年硕士论文

【摘要】：随着电子通讯行业的快速发展,中国各大运营商之间的竞争愈加激烈。目前中国主要有三大运营商：移动、联通、电信。三大运营商都有自己的客户群,但同时都面临客户流失率攀升的问题,其主要原因是现有的解决方案没有运用数据挖掘理论基础,以及现有系统大多以某个城市作为试点进行设计,缺乏通用性和拓展性,其渠道也未全面扩展。针对上述问题,存量维系系统运用数据挖掘的决策树方法建立客户群模型,并设计互斥矩阵规则、优惠控制规则,从而实现针对性营销,切实提高客户维系率。具体来说,主要工作包括以下四个方面：1、设计存量维系系统总体架构,存量维系系统不是一个单一的系统,需要时刻与其它系统进行交互。存量维系系统从地市中心获取基础数据,为营销推荐提供客户基本属性信息；存量维系系统从标签库系统中提取客户群数据,然后将数据推送到存量维系系统平台；在经营分析系统中提供统一登录入口,即从经营分析系统中可以登录到存量维系系统平台。2、研究客户群模型。存量维系系统采用数据挖掘决策树方法,基于ID3算法创建属性结点；利用客户基本属性进行分类,建立客户群模型。3、设计政策推荐方法。利用互斥矩阵规则对将要推荐的政策进行过滤,如果政策之间互斥或者已经推荐过,则不推荐,反之则推荐；利用优惠控制规则可有效避免个人客户和集团客户享受过度的优惠,让重复优惠控制的模糊边界不复存在。4、实现存量维系系统。设计存量维系系统各个功能模块；整合客户群模型技术和政策推荐方法；研究各大渠道对存量维系系统的作用和影响；在和其它系统进行交互的过程中,采用Web Service技术有效提高实时性和安全性；最终完成存量维系系统的开发,并进行测试验证。综上所述,本论文设计并实现了存量维系系统,该系统运用数据挖掘技术建立客户群模型,在客户群模型基础上设计政策推荐方法,并对将要推荐的政策进行互斥矩阵规则和优惠控制规则过滤。最终达到提高客户维系效率,最大程度的挽留客户,使客户保持长期的稳定；并使移动切实进入B2C (Business-to-Customer)互联网电子商务阶段。
[Abstract]:With the rapid development of the electronic communications industry, the competition among the major operators in China has become increasingly fierce. At present, there are three major operators in China: Mobile, Unicom, telecommunications. The three major operators have their own customers, but both are facing the problem of rising customer loss rate. The main reason is that the existing solutions do not use data digging. The theoretical basis of mining and most of the existing systems are designed in a city as a pilot project, lack of generality and expansibility, and the channel has not been expanded. In view of the above problems, the stock maintenance system uses the decision tree method of data mining to establish the customer group model, and designs the mutual exclusion matrix rules and preferential control rules, thus realizing the pertinence. Marketing, and effectively improve the customer maintenance rate. Specifically, the main work includes the following four aspects: 1, design stock to maintain the overall framework of the system, the stock maintenance system is not a single system, the need to interact with other systems. Stock maintenance system from the center of the city to obtain basic data for marketing recommendations to provide basic customers The stock maintenance system extracts the customer group data from the tag library system, and then pushes the data to the stock maintenance system platform, and provides the unified login entry in the management analysis system, that is, we can log in to the stock maintenance system platform.2 from the management analysis system and study the customer group model. The stock maintenance system adopts the data. Mining decision tree method, based on ID3 algorithm to create attribute nodes; use customer basic attributes to classify, establish customer group model.3, design policy recommendation method. Use mutual exclusion matrix rules to filter the policy that will be recommended, if policy mutual exclusion or already recommended, it is not recommended, and vice versa; use preferential control. The rules of the system can effectively avoid the excessive preferences of individual customers and group customers, let the fuzzy boundaries of the repeated preferential control do not exist in the existence of.4 and realize the stock maintenance system. The design stock maintains each function module of the system, integrates the customer group model technology and the policy recommendation method, and studies the role and shadow of the large channels to the stock maintenance system. In the process of interaction with other systems, the Web Service technology is used to effectively improve the real-time and security; finally, the stock maintenance system is developed and tested. In summary, this paper designs and implements a stock maintenance system. The system uses data mining technology to establish a customer group model, in the customer group model. On the basis of the model, we design policy recommendation methods and filter the mutually exclusive matrix rules and preferential control rules for the policies to be recommended. Finally, it can improve the customer maintenance efficiency, maximize the retention of customers, keep the customers stable for a long time, and make the move into the B2C (Business-to-Customer) Internet e-commerce phase.
【学位授予单位】：东南大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP311.13

【参考文献】