数据挖掘在通信行业中高端用户保有的应用
|
摘要
信息产业部为了全业务发展的需求,对通信行业的六家运营商进行第四次重组,新电信和新联通凭借全业务均衡发展的优势在新形势下的竞争中开展激烈的用户争夺战,重组后的中国移动与中国铁通的联合短板非常明显,但原有的移动业务绝对优势也是不容小觑,这给重组后的市场竞争增添了不少变数。原有的存量客户是三家运营商必争的资源,尤其是优质的中高端客户的争夺更是硝烟弥漫。在这种竞争态势下,如何防止中高端客户流失,有效的保有中高端客户是摆在几家运营商面前的一道难题,也是当前发展市场、维系客户工作的重中之重。利用数据挖掘技术对中高端客户的流失进行预警,从而确保在流失前保留住客户。
数据挖掘技术就是从海量的数据中抽取隐含其中潜在的、不为人知的有用信息、模式或趋势。本文的目的就是研究这种技术,并将这种技术应用到通信企业的中高端预警模型中,实现对中高端客户的有效保有。
本文以某通信企业的历史数据为研究对象,基于数据挖掘技术,建立客户流失预警模型,采用的决策树分析方法进行数据挖掘分析,有利于电信企业在海量数据中的挖掘分析。论文的主要内容如下:
1、介绍数据挖掘技术和数据仓库的基本理论知识,并介绍数据挖掘的过程以及几种常见的工具和算法。
2、研究数据仓库的设计过程,知道如何建立一个数据仓库模型的关键步骤数据准备包括:数据抽取、数据清洗、数据转换、数据加载和数据修正五个步骤。
3、在通信行业中,一般可用聚类方法来进行客户的细分。而针对客户流失预测及客户保有问题,多会采用决策树方法,相对来说,决策树方法的其结构和推理的过程更清楚。研究决策树算法,根据算法中的不足进行改进算法,采用决策树的优化算法,能够达到提高效率同时又能保证数据质量,在同类分析算法中相对较优。
4、介绍数据仓库的设计和构建,建立预测模型的过程采取六个步骤进行:业务分析、数据准备、变量选择、模型构建、模型评估和模型部署。
5、基于前面的分析建立预测模型,利用某移动公司的用户样本数据,对系统进行了验证,通过大量数据处理和反复试验得到了比较理想的结果,并以此为基础制定了相应的营销方案,对执行中高端用户的保有工作起到了非常重要的作用。
本论文提出的基于决策树的客户流失预警方案可以为通信企业在中高端用户的保有工作和方案上提供一定参考,有一定的应用价值。
Ministry of Information Industry, the demand for all business development, six of the
communications industry operators in the fourth reorganization, the new telecommunications and the new China Unicom with the advantages of the balanced development of all business in the new situation to carry out under the fierce competition in the battle for customers, restructured China Mobile and China Railcom's joint short board is very obvious, but the original advantages of mobile services is an absolute can not be underestimated, which after the reorganization of competition in the market to add a lot of variables. The stock of existing customers are the three carriers compete for resources, particularly in high-quality competition for high-end customers is smoke. In this competitive situation, how to prevent the loss of high-end customers, effective to maintain high-end customers are placed in front of several operators a difficult problem, but also the current top priority. Using data mining technology to the loss of high-end early warning customers to ensure that the loss of the former retain customers.
The data mining technology is abstracted from huge data of implicit potential, unknown useful information, pattern or trends. The purpose of this paper is to research this kind of technology, and will this technology is applied to the communication enterprises in high-end warning model, to achieve the effective retain in high-end customers. Taking a historical data communications company, as the research object, based on data mining technology, a customer churn prediction model, using the decision tree analysis methods for data mining analysis, more use of telecommunications companies in the mining massive data analysis, the paper the main content as follows:
1, describes data mining and data warehousing knowledge of the basic theory and describes the process of data mining as well as several common tools and algorithms.
2, the data warehouse design process, know how to build a data warehouse model is a key step in data preparation, including: data extraction, data cleansing, data transformation, data loading and data correction five steps.
3, in the communications industry in general, clustering methods can be used for customer segmentation. The customer churn prediction and customer retention issues, and more will use the decision tree method, relatively speaking, decision tree structure and its reasoning process more clearly. Of decision tree, according to algorithm deficiencies in the improved algorithm, using the decision tree algorithm, can achieve greater efficiency while ensuring data quality, analysis of algorithms in the same relatively better.
4 introduces the data warehouse design and build, build the prediction model of the process to take six steps: business analysis, data preparation, variable selection, model building, model evaluation and model deployment.
5, based on the previous analysis to establish prediction model, the user of a mobile company sample data, the system was verified repeatedly by a large number of data processing and test results have been ideal, and as a basis to develop the corresponding marketing programs on the implementation of the tenure of the work in the high-end users play a very important role.
This paper presents a decision tree based on the customer churn warning scheme for the communication enterprises in May in high-end can keep on working and scheme provides certain reference, have certain application value
数据挖掘技术就是从海量的数据中抽取隐含其中潜在的、不为人知的有用信息、模式或趋势。本文的目的就是研究这种技术,并将这种技术应用到通信企业的中高端预警模型中,实现对中高端客户的有效保有。
本文以某通信企业的历史数据为研究对象,基于数据挖掘技术,建立客户流失预警模型,采用的决策树分析方法进行数据挖掘分析,有利于电信企业在海量数据中的挖掘分析。论文的主要内容如下:
1、介绍数据挖掘技术和数据仓库的基本理论知识,并介绍数据挖掘的过程以及几种常见的工具和算法。
2、研究数据仓库的设计过程,知道如何建立一个数据仓库模型的关键步骤数据准备包括:数据抽取、数据清洗、数据转换、数据加载和数据修正五个步骤。
3、在通信行业中,一般可用聚类方法来进行客户的细分。而针对客户流失预测及客户保有问题,多会采用决策树方法,相对来说,决策树方法的其结构和推理的过程更清楚。研究决策树算法,根据算法中的不足进行改进算法,采用决策树的优化算法,能够达到提高效率同时又能保证数据质量,在同类分析算法中相对较优。
4、介绍数据仓库的设计和构建,建立预测模型的过程采取六个步骤进行:业务分析、数据准备、变量选择、模型构建、模型评估和模型部署。
5、基于前面的分析建立预测模型,利用某移动公司的用户样本数据,对系统进行了验证,通过大量数据处理和反复试验得到了比较理想的结果,并以此为基础制定了相应的营销方案,对执行中高端用户的保有工作起到了非常重要的作用。
本论文提出的基于决策树的客户流失预警方案可以为通信企业在中高端用户的保有工作和方案上提供一定参考,有一定的应用价值。
Ministry of Information Industry, the demand for all business development, six of the
communications industry operators in the fourth reorganization, the new telecommunications and the new China Unicom with the advantages of the balanced development of all business in the new situation to carry out under the fierce competition in the battle for customers, restructured China Mobile and China Railcom's joint short board is very obvious, but the original advantages of mobile services is an absolute can not be underestimated, which after the reorganization of competition in the market to add a lot of variables. The stock of existing customers are the three carriers compete for resources, particularly in high-quality competition for high-end customers is smoke. In this competitive situation, how to prevent the loss of high-end customers, effective to maintain high-end customers are placed in front of several operators a difficult problem, but also the current top priority. Using data mining technology to the loss of high-end early warning customers to ensure that the loss of the former retain customers.
The data mining technology is abstracted from huge data of implicit potential, unknown useful information, pattern or trends. The purpose of this paper is to research this kind of technology, and will this technology is applied to the communication enterprises in high-end warning model, to achieve the effective retain in high-end customers. Taking a historical data communications company, as the research object, based on data mining technology, a customer churn prediction model, using the decision tree analysis methods for data mining analysis, more use of telecommunications companies in the mining massive data analysis, the paper the main content as follows:
1, describes data mining and data warehousing knowledge of the basic theory and describes the process of data mining as well as several common tools and algorithms.
2, the data warehouse design process, know how to build a data warehouse model is a key step in data preparation, including: data extraction, data cleansing, data transformation, data loading and data correction five steps.
3, in the communications industry in general, clustering methods can be used for customer segmentation. The customer churn prediction and customer retention issues, and more will use the decision tree method, relatively speaking, decision tree structure and its reasoning process more clearly. Of decision tree, according to algorithm deficiencies in the improved algorithm, using the decision tree algorithm, can achieve greater efficiency while ensuring data quality, analysis of algorithms in the same relatively better.
4 introduces the data warehouse design and build, build the prediction model of the process to take six steps: business analysis, data preparation, variable selection, model building, model evaluation and model deployment.
5, based on the previous analysis to establish prediction model, the user of a mobile company sample data, the system was verified repeatedly by a large number of data processing and test results have been ideal, and as a basis to develop the corresponding marketing programs on the implementation of the tenure of the work in the high-end users play a very important role.
This paper presents a decision tree based on the customer churn warning scheme for the communication enterprises in May in high-end can keep on working and scheme provides certain reference, have certain application value
引文
[1]邵峰晶,于忠清等.数据挖掘原理与算法(第二版).北京:科学出版社,2009
[2]李爱群,乔晗等.基于分布式混合数据挖掘的电信客户流失分析.计算机技术与发展,2010.10第20卷.第10期
[3]董昊.模糊聚类分析法在房地产企业分类中的应用[J].合作经济与科技,2010(1)
[4]史小松,黄勇杰,刘永革.数据挖掘技术中聚类的几种常用方法比较[J].中国科技信息,2009(20)
[5]陈京民.数据仓库与数据挖掘技术.北京:电子工业出版社,2002.12～15
[6] Jiawei Han,Micheline Kamber .数据挖掘概念与技术.范明,孟小峰等译.机械工业出版社,2007
[7] Raymond T.Ng and JiaweiHan.CLARANS:A Method for C lustering 0 bjects for K-means clustering [J].Pattem Recognition Letters,2004(25):1293-1302
[8]梁循.数据挖掘算法与应用[M].北京大学出版社,2002
[9]宋艳,张炎欣.数据挖掘技术在CRM中的应用.企业导报.2009.(4)
[10]王维佳,缪柏其,魏国省.数据挖掘—电信客户流失分析预测[J]数理统计与管理,2006.25(4)419
[11] Stanley A.Brown. Customer Relationship Management:A Strategic Imperative in the world of E-Business.John wiley&Sons,2000:6-26
[12]邹丽珊,郑金华.数据聚类的共同进化方法.计算机工程与应用,2004.18
[13] John shawetaylor,Keith Howkerand PeterBurge.Detection of Fraud in Bobile Telecommunition.Information Security Technial Report,2002,4(1):202-203
[14]汤小文,蔡庆生.数据挖掘在电信业中的应用.计算机工程,2004,30(6):36-37
[15] heng S M,oh S Y.Support vector machines with binary tree architecture for multi class classitication.Neural Information Processing Letters and Reviews,2004,2(3)
[16]陈黎,黄心汉,王敏.基于聚类分析的车牌字符分割方法.北京:计算机工程与应用,2002.6:211～212
[17]丁秋林,力士奇.客户关系管理[M].北京:清华大学出版社,2002.32～37
[18] Alex Berson,Stephen Smith ,Kurt Thearling.构建面向CRM的数据挖掘应用.贺奇,郑岩等译.北京:人民邮电出版社,2001.
[19]杨瑞桢,黄传武,电信市场营销基本理论与实务.北京:北京邮电大学出版社,2003.10
[20]刘世平,姚玉辉博士.数据挖掘工具的评判.上海:数字财富,2003,6
[21]杨树莲.数据挖掘在电信行业客户流失分析中的应用.计算机与现代,2005(2)
[22] Kim Hee Su,Yoon Choong Han.Determinations of Subecriber Churn and Customer Loyalty in the Korean Mobile Telephony Market. Telecommunications Policy,2004,28(9)
[23]仇春芳,李卫卫.数据挖掘在电信客户流失分析中的应用.通信世界,2007,2(12):18-21
[24]郭明,郑惠莉.用数据挖掘法分析电信客户流失,现代通信,2005(3):7-9
[25] Deloitte Consultiong.四川移动客户细分模型项目报告.北京:德勤,2004
[26]夏国金,金炜东,客户流失预测中两类错误的平衡控制研究.营销科学学报,2006,2(4):1-7
[27] KriegelH.P.,Pfeifle M .Dendity-based clustering of uncertain data[C].Proc.11 th Int.Conf.on knowledge Discovery and Data Mining (KDD’05),chicago,IL,2005,672-677
[28]于晨捷,袁晓洁,马涛.数据挖掘中趋势模型的建立与分析,北京:计算机工程与应用,2002.8 198-200
[29]田永青,邢庆国,朱仲英.一种基于税务系统数据仓库的模糊数据挖掘算法的研究.计算机工程与应用,2002.10: 206-208
[30] Melody Y.Kiang,Michael Y.hu,Dorothy M. Fisher .An Extended Self-organizing Map Network for Market Segmentation—A Telecommunication Example[J].Decision Support ystems 2006,6:36-47.
[31] Stone, M. et al,Database marketing and customer recruitment, retention and development: what is the technological state of the art? Journal of Database Marketing 1998,5(4),303-331.
[32]谭建豪.数据挖掘技术[M].北京:中国水利电出版社,2009.
[33]贝尔森,史密斯,西瑞林.构建面向CRM的数据挖掘应用[M].贺奇,译.北京:人民邮电出版社,2001
[34]闭小梅,闭瑞华.KNN分类算法研究[J].科技创新导报,2009,(14):31
[35]陈明,何书萍,李凡长.一种李群机器学习线性分类算法研究[J].微电子学与计算机,2009,(10):170-173
[36] Christine Normile. Business Intelligence for the Telecommunications Industry[J]. Ingres Corporation, 2008.
[37]路红梅.基于决策树的经典算法综述[J].宿州学院学报,2007(4):91-95
[38] AITKENHEAD M J. Aco-evolving decision tree classification method[J].expert System with Application ,2008(34):18-25
[39] OLARU C,WEHENKEL L.A complete fuzzy decision tree technique [J].Fuzzy Sets and Systems,2003,138(2):221-254
[40]史忠植.知识发现[M].北京:清华大学出版社,2002
[41] HAN J.W,KAMBER M.Data Mining Concepts and Techniques[M].Morgan Kaufmann,2001
[42]李雄飞,李军.Data Mining and Knowledge Discovery.数据挖掘与知识发现[M].北京高等教育出版社,2003
[43]王玮,蔡莲红,关联规则的高效挖掘算法研究[J].小型微型计算机系统,2002,48-52
[44]薛薇,王益锋,赵璋.基于客户细分的电信客户流失防范对策研究[J].经理理论研究,2007(4):48-50
[45] Salvatore Ruggieri.Efficient C4.5[J].IEEE Transaction on Knowledge and Data Engineering 2002,14(2):438-444
[46]李宁,乐琦.决策树算法及常见问题的解决[J].计算机与数字工程,2005,(3)
[47]陈文伟,黄金才等.数据仓库与数据挖掘[M].北京:人民邮电出版社,2004:1
[48]毛国君等.数据挖掘原理与算法[M].北京:清华大学出版社,2005:7
[49] Paplo Giudici.实用数据挖掘[M].北京:电子工业出版社,2004
[50]谭旭等.利用决策树发掘分类规则的算法研究[J].云南大学学报,2000,22(6)
[51] PANG-NING TAN,MICHAEL STEINBACH,VIPIN KUMAR.数据挖掘导论[M].范明,范宏建,等,译.北京:人民邮电出版社,2006:2-3
[52] D Fournier,B Cremilleux.A quality index for decision tree pruning[J]. Knowledge-based Systems,2002,15(1):37-43
[53]杨静,张楠男,李建等.决策树算法的研究与应用[J].计算机技术与发展,2010(2)
[2]李爱群,乔晗等.基于分布式混合数据挖掘的电信客户流失分析.计算机技术与发展,2010.10第20卷.第10期
[3]董昊.模糊聚类分析法在房地产企业分类中的应用[J].合作经济与科技,2010(1)
[4]史小松,黄勇杰,刘永革.数据挖掘技术中聚类的几种常用方法比较[J].中国科技信息,2009(20)
[5]陈京民.数据仓库与数据挖掘技术.北京:电子工业出版社,2002.12～15
[6] Jiawei Han,Micheline Kamber .数据挖掘概念与技术.范明,孟小峰等译.机械工业出版社,2007
[7] Raymond T.Ng and JiaweiHan.CLARANS:A Method for C lustering 0 bjects for K-means clustering [J].Pattem Recognition Letters,2004(25):1293-1302
[8]梁循.数据挖掘算法与应用[M].北京大学出版社,2002
[9]宋艳,张炎欣.数据挖掘技术在CRM中的应用.企业导报.2009.(4)
[10]王维佳,缪柏其,魏国省.数据挖掘—电信客户流失分析预测[J]数理统计与管理,2006.25(4)419
[11] Stanley A.Brown. Customer Relationship Management:A Strategic Imperative in the world of E-Business.John wiley&Sons,2000:6-26
[12]邹丽珊,郑金华.数据聚类的共同进化方法.计算机工程与应用,2004.18
[13] John shawetaylor,Keith Howkerand PeterBurge.Detection of Fraud in Bobile Telecommunition.Information Security Technial Report,2002,4(1):202-203
[14]汤小文,蔡庆生.数据挖掘在电信业中的应用.计算机工程,2004,30(6):36-37
[15] heng S M,oh S Y.Support vector machines with binary tree architecture for multi class classitication.Neural Information Processing Letters and Reviews,2004,2(3)
[16]陈黎,黄心汉,王敏.基于聚类分析的车牌字符分割方法.北京:计算机工程与应用,2002.6:211～212
[17]丁秋林,力士奇.客户关系管理[M].北京:清华大学出版社,2002.32～37
[18] Alex Berson,Stephen Smith ,Kurt Thearling.构建面向CRM的数据挖掘应用.贺奇,郑岩等译.北京:人民邮电出版社,2001.
[19]杨瑞桢,黄传武,电信市场营销基本理论与实务.北京:北京邮电大学出版社,2003.10
[20]刘世平,姚玉辉博士.数据挖掘工具的评判.上海:数字财富,2003,6
[21]杨树莲.数据挖掘在电信行业客户流失分析中的应用.计算机与现代,2005(2)
[22] Kim Hee Su,Yoon Choong Han.Determinations of Subecriber Churn and Customer Loyalty in the Korean Mobile Telephony Market. Telecommunications Policy,2004,28(9)
[23]仇春芳,李卫卫.数据挖掘在电信客户流失分析中的应用.通信世界,2007,2(12):18-21
[24]郭明,郑惠莉.用数据挖掘法分析电信客户流失,现代通信,2005(3):7-9
[25] Deloitte Consultiong.四川移动客户细分模型项目报告.北京:德勤,2004
[26]夏国金,金炜东,客户流失预测中两类错误的平衡控制研究.营销科学学报,2006,2(4):1-7
[27] KriegelH.P.,Pfeifle M .Dendity-based clustering of uncertain data[C].Proc.11 th Int.Conf.on knowledge Discovery and Data Mining (KDD’05),chicago,IL,2005,672-677
[28]于晨捷,袁晓洁,马涛.数据挖掘中趋势模型的建立与分析,北京:计算机工程与应用,2002.8 198-200
[29]田永青,邢庆国,朱仲英.一种基于税务系统数据仓库的模糊数据挖掘算法的研究.计算机工程与应用,2002.10: 206-208
[30] Melody Y.Kiang,Michael Y.hu,Dorothy M. Fisher .An Extended Self-organizing Map Network for Market Segmentation—A Telecommunication Example[J].Decision Support ystems 2006,6:36-47.
[31] Stone, M. et al,Database marketing and customer recruitment, retention and development: what is the technological state of the art? Journal of Database Marketing 1998,5(4),303-331.
[32]谭建豪.数据挖掘技术[M].北京:中国水利电出版社,2009.
[33]贝尔森,史密斯,西瑞林.构建面向CRM的数据挖掘应用[M].贺奇,译.北京:人民邮电出版社,2001
[34]闭小梅,闭瑞华.KNN分类算法研究[J].科技创新导报,2009,(14):31
[35]陈明,何书萍,李凡长.一种李群机器学习线性分类算法研究[J].微电子学与计算机,2009,(10):170-173
[36] Christine Normile. Business Intelligence for the Telecommunications Industry[J]. Ingres Corporation, 2008.
[37]路红梅.基于决策树的经典算法综述[J].宿州学院学报,2007(4):91-95
[38] AITKENHEAD M J. Aco-evolving decision tree classification method[J].expert System with Application ,2008(34):18-25
[39] OLARU C,WEHENKEL L.A complete fuzzy decision tree technique [J].Fuzzy Sets and Systems,2003,138(2):221-254
[40]史忠植.知识发现[M].北京:清华大学出版社,2002
[41] HAN J.W,KAMBER M.Data Mining Concepts and Techniques[M].Morgan Kaufmann,2001
[42]李雄飞,李军.Data Mining and Knowledge Discovery.数据挖掘与知识发现[M].北京高等教育出版社,2003
[43]王玮,蔡莲红,关联规则的高效挖掘算法研究[J].小型微型计算机系统,2002,48-52
[44]薛薇,王益锋,赵璋.基于客户细分的电信客户流失防范对策研究[J].经理理论研究,2007(4):48-50
[45] Salvatore Ruggieri.Efficient C4.5[J].IEEE Transaction on Knowledge and Data Engineering 2002,14(2):438-444
[46]李宁,乐琦.决策树算法及常见问题的解决[J].计算机与数字工程,2005,(3)
[47]陈文伟,黄金才等.数据仓库与数据挖掘[M].北京:人民邮电出版社,2004:1
[48]毛国君等.数据挖掘原理与算法[M].北京:清华大学出版社,2005:7
[49] Paplo Giudici.实用数据挖掘[M].北京:电子工业出版社,2004
[50]谭旭等.利用决策树发掘分类规则的算法研究[J].云南大学学报,2000,22(6)
[51] PANG-NING TAN,MICHAEL STEINBACH,VIPIN KUMAR.数据挖掘导论[M].范明,范宏建,等,译.北京:人民邮电出版社,2006:2-3
[52] D Fournier,B Cremilleux.A quality index for decision tree pruning[J]. Knowledge-based Systems,2002,15(1):37-43
[53]杨静,张楠男,李建等.决策树算法的研究与应用[J].计算机技术与发展,2010(2)
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |