本文選題：存量維系 + 決策樹��； 參考：《東南大學(xué)》2015年碩士論文

【摘要】：隨著電子通訊行業(yè)的快速發(fā)展,中國(guó)各大運(yùn)營(yíng)商之間的競(jìng)爭(zhēng)愈加激烈。目前中國(guó)主要有三大運(yùn)營(yíng)商：移動(dòng)、聯(lián)通、電信。三大運(yùn)營(yíng)商都有自己的客戶群,但同時(shí)都面臨客戶流失率攀升的問(wèn)題,其主要原因是現(xiàn)有的解決方案沒(méi)有運(yùn)用數(shù)據(jù)挖掘理論基礎(chǔ),以及現(xiàn)有系統(tǒng)大多以某個(gè)城市作為試點(diǎn)進(jìn)行設(shè)計(jì),缺乏通用性和拓展性,其渠道也未全面擴(kuò)展。針對(duì)上述問(wèn)題,存量維系系統(tǒng)運(yùn)用數(shù)據(jù)挖掘的決策樹方法建立客戶群模型,并設(shè)計(jì)互斥矩陣規(guī)則、優(yōu)惠控制規(guī)則,從而實(shí)現(xiàn)針對(duì)性營(yíng)銷,切實(shí)提高客戶維系率。具體來(lái)說(shuō),主要工作包括以下四個(gè)方面：1、設(shè)計(jì)存量維系系統(tǒng)總體架構(gòu),存量維系系統(tǒng)不是一個(gè)單一的系統(tǒng),需要時(shí)刻與其它系統(tǒng)進(jìn)行交互。存量維系系統(tǒng)從地市中心獲取基礎(chǔ)數(shù)據(jù),為營(yíng)銷推薦提供客戶基本屬性信息；存量維系系統(tǒng)從標(biāo)簽庫(kù)系統(tǒng)中提取客戶群數(shù)據(jù),然后將數(shù)據(jù)推送到存量維系系統(tǒng)平臺(tái)；在經(jīng)營(yíng)分析系統(tǒng)中提供統(tǒng)一登錄入口,即從經(jīng)營(yíng)分析系統(tǒng)中可以登錄到存量維系系統(tǒng)平臺(tái)。2、研究客戶群模型。存量維系系統(tǒng)采用數(shù)據(jù)挖掘決策樹方法,基于ID3算法創(chuàng)建屬性結(jié)點(diǎn)；利用客戶基本屬性進(jìn)行分類,建立客戶群模型。3、設(shè)計(jì)政策推薦方法。利用互斥矩陣規(guī)則對(duì)將要推薦的政策進(jìn)行過(guò)濾,如果政策之間互斥或者已經(jīng)推薦過(guò),則不推薦,反之則推薦；利用優(yōu)惠控制規(guī)則可有效避免個(gè)人客戶和集團(tuán)客戶享受過(guò)度的優(yōu)惠,讓重復(fù)優(yōu)惠控制的模糊邊界不復(fù)存在。4、實(shí)現(xiàn)存量維系系統(tǒng)。設(shè)計(jì)存量維系系統(tǒng)各個(gè)功能模塊；整合客戶群模型技術(shù)和政策推薦方法；研究各大渠道對(duì)存量維系系統(tǒng)的作用和影響；在和其它系統(tǒng)進(jìn)行交互的過(guò)程中,采用Web Service技術(shù)有效提高實(shí)時(shí)性和安全性；最終完成存量維系系統(tǒng)的開發(fā),并進(jìn)行測(cè)試驗(yàn)證。綜上所述,本論文設(shè)計(jì)并實(shí)現(xiàn)了存量維系系統(tǒng),該系統(tǒng)運(yùn)用數(shù)據(jù)挖掘技術(shù)建立客戶群模型,在客戶群模型基礎(chǔ)上設(shè)計(jì)政策推薦方法,并對(duì)將要推薦的政策進(jìn)行互斥矩陣規(guī)則和優(yōu)惠控制規(guī)則過(guò)濾。最終達(dá)到提高客戶維系效率,最大程度的挽留客戶,使客戶保持長(zhǎng)期的穩(wěn)定；并使移動(dòng)切實(shí)進(jìn)入B2C (Business-to-Customer)互聯(lián)網(wǎng)電子商務(wù)階段。
[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.
【學(xué)位授予單位】：東南大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2015
【分類號(hào)】：TP311.13

【參考文獻(xiàn)】

相關(guān)期刊論文 前1條

1 吳茂昌;陽(yáng)玉琴;;基于MVC模式的Java主流框架整合技術(shù)研究[J];計(jì)算機(jī)與數(shù)字工程;2009年10期

，

本文編號(hào)：1956700