基于数据挖掘技术的电信客户流失预测模型的研究与应用
|
摘要
数据挖掘技术是利用已知的数据通过建立数学模型的方法找出隐含的业务规则。在国内随着对数据挖掘技术的重视,数据挖掘技术的应用研究也越来越广,其中对电信行业的客户流失分析就是一大热点。客户流失分析是通过对以往流失客户的历史数据进行分析,找出可能离网用户的特征,及时采取相应的措施,减少客户流失的发生。这对企业降低运营成本,提高经营业绩有着极为重要的意义。
本文从提高数据挖掘的效率和精度的目的出发,对BP神经网络预测模型进行了有益的改善,同时给出了基于粗糙集理论的属性约简和BP神经网络相结合的客户流失预测方法。通过属性约简技术对神经网络的输入属性空间进行约简,采用神经网络对约简后的数据进行挖掘。此方法充分发挥了粗糙集理论在约简知识方面的能力和神经网络预测精度高的特点,应用于电信客户流失预测技术研究中,取得了较好的效果。
在上述研究的基础上,本文根据数据挖掘建模过程建立电信客户流失预测模型,给出电信客户流失行为预测的解决方案。并对预测模型进行性能评估。评估结果表明本文建立的预测模型是可行的。本文构建的预测模型对解决电信客户流失预测方面的问题具有应用价值。
Data mining technology makes use of existed data to find out the underling business rule by establishing mathematical model. The prediction of customer churn in telecommunication industry is very important. The prediction of customer churn is to analyze the churned customer's historical data .So that the reasons why they left might be found out. It will help the telecom company to adopt measures early to reduce customer churn. It has a very important significance for enterprises to reduce operating costs and enhance operating performance.
In order to improve the efficiency of data mining, the thesis proposes a data mining method based on rough set and artificial neural network. By reduction processing to the import space, this method adopts artificial neural network for data mining on the reduced training data. The method exerts the ability of rough set's reduction knowledge and the high precision feature of artificial neural network. It gains a very good result to apply this method to the prediction of telecom customer churn.
The thesis builds a telecom customer churn prediction model, guided by the above data mining techniques. And a prototype is implemented and evaluated on real data. The results of evaluation show that the prediction model is feasible. The prediction model helps to predict customer-churn behavior in the telecommunication industries.
本文从提高数据挖掘的效率和精度的目的出发,对BP神经网络预测模型进行了有益的改善,同时给出了基于粗糙集理论的属性约简和BP神经网络相结合的客户流失预测方法。通过属性约简技术对神经网络的输入属性空间进行约简,采用神经网络对约简后的数据进行挖掘。此方法充分发挥了粗糙集理论在约简知识方面的能力和神经网络预测精度高的特点,应用于电信客户流失预测技术研究中,取得了较好的效果。
在上述研究的基础上,本文根据数据挖掘建模过程建立电信客户流失预测模型,给出电信客户流失行为预测的解决方案。并对预测模型进行性能评估。评估结果表明本文建立的预测模型是可行的。本文构建的预测模型对解决电信客户流失预测方面的问题具有应用价值。
Data mining technology makes use of existed data to find out the underling business rule by establishing mathematical model. The prediction of customer churn in telecommunication industry is very important. The prediction of customer churn is to analyze the churned customer's historical data .So that the reasons why they left might be found out. It will help the telecom company to adopt measures early to reduce customer churn. It has a very important significance for enterprises to reduce operating costs and enhance operating performance.
In order to improve the efficiency of data mining, the thesis proposes a data mining method based on rough set and artificial neural network. By reduction processing to the import space, this method adopts artificial neural network for data mining on the reduced training data. The method exerts the ability of rough set's reduction knowledge and the high precision feature of artificial neural network. It gains a very good result to apply this method to the prediction of telecom customer churn.
The thesis builds a telecom customer churn prediction model, guided by the above data mining techniques. And a prototype is implemented and evaluated on real data. The results of evaluation show that the prediction model is feasible. The prediction model helps to predict customer-churn behavior in the telecommunication industries.
引文
[1]叶松云.我国电信行业客户流失管理的建模、分析及应用研究[D].厦门:暨 南 大 学硕士学位论文,2004.
[2]盛昭瀚、柳炳祥.客户流失危机分析的决策树方法[J].管理科学学报,2005,8(2):20-25
[3]贾琳、李明.基于数据挖掘的电信客户流失模型的建立与实现[J].计算机工程与应用,2004,40(4):185-187
[4]HUNGSY , YENDC , WANGHY. Applying data mining to telecom churn management[J].Expert Systems with Applications,2006,31(3):515-524.
[5]Mozer Mc. Predicting subscriber Dissatisfaction and Improving Retention in the Wireless Telecommunication Industry[J].IEEE Trans on Neural Networks, 2000,11(3)
[6]Lian Yan. Predicting customer behavior in telecommunications [J].Intelligent Systems,IEEE,2004(2)
[7]Masand B,Datta P,Mani DR.CHAMP:A Prototype for Automated Celluar Churn Prediction[J].Data Mining and Knowledge Discovery,1999(3)
[8]王平.利用数据挖掘实现电信业的客户流失预测分析[D].西南交通大学硕士论文, 2003 .
[9]郭明,郑惠莉,卢毓伟.基于贝叶斯网络的客户流失分析[J].南京邮电大学学报:自然科学版,2005,25(5):79-83
[10]陈文伟,黄金才.数据仓库与数据挖掘[M].北京:人民邮电出版社,2004,5-7
[11] R.Agrawal and R.Srikant, "Fast Algorithms for Mining Association Rules", In Proc. of the 20th Int'1 Conference on Very Large Databases. Santiago, Chile, 1994.
[12] 史忠植.知识发现[M].清华大学出版社,2002.
[13] 陈京民.数据仓库与数据挖掘技术[M].电子工业出版社,2002,1:25-26
[14] CRI SP-DM1.0数据挖掘方法论指南,第1版.CRISP-DM协会.2000.
[15] 耿晓中,张冬梅.数据挖掘综述[J].长春师范学院学报(自然科学版),2006,25 (3):24-27
[16] H.Mannila , H.Toivonen and LVerkamo.Efficient Algorithms for Discovering Association Rules.In AAAI Workshop, Knowledge Discovery in Databases, 181-192, 1994.
[17]Pawlak Z. Rough Set Theoretical Aspects of Reasoning about Data[M].Dordrecht : Kluwer Academic Publishers ,1991.
[18] 龙志勇.数据挖掘在电信行业客户关系管理中的应用[J].信系网络,2003 (12):24-26
[19] Pawlak Z. Rough Sets[J] . International Journal of Computer and Information Science ,1982 , (5) :341 356
[20]石红,沈毅,刘志言等.关于粗糙集理论及应用问题的研究[J].计算机工程,2003,29(3):1-4
[21] Pawlak Z et al.Rough sets[J].Communications of ACM,1995,38(11):89-95.
[22] 刘清.Rough集及Rough推理[M]. 北京: 科学出版社, 2001.
[23] 王国胤.Rough集理论与知识获取[M].西安: 西安交通大学出版社,2001.
[24] 张文修,吴伟志等.粗糙集理论与方法[M].北京: 科学出版社,2001.
[25] 常梨云,王国胤,吴渝. 一种基于Rough Set 理论的属性约简及规则提取方法[J] . 软件学报,1999 ,10 (11) :1206-1211.
[26]芦晓红,陈世权,吴今培.基于可辨识矩阵的启发式属性约简算法及其应用[J].计算机工程,2003,29(1) :56-59.
[27]Hu X H .Knowledge discovery in databases: An attribute-oriented rough s et approach. Canada:University of Regina Pr.,1995.
[28]Hu X H ,CerconeN .Learning in relational databases:A rough set approach. In t.J.of Computational Intelligence,1995,11(2):323-338.
[29]Jakub W. Genetic algorithms in decomposition and classification problems. In:L.Polkows,A.Skowron(eds.):Rough Sets in Knowledge Discovery.Physica-Verlag,Heidelberg,1998(2):471-487
[30] Kryszkiewicz M, Rybinski H. Finding reducts in composed information systems. Int Work shop on Rough Sets and Knowledge Discovery RSKD'93,London:SpringerVerlag,1993:259-268.
[31] Starzyk J, Nelson D E, Sturtz K. Reducts in composed information systems. Bulletin of International Rough Set Society,1999,3(1-2):19-22.
[32] Starzyk J, Nelson D E, Sturtz K. Reducts. A mathematical foundation for improved reduct generation in informations ystems. Journal of Knowledge and Information Systems,2000,2(2):131-146 [33 ] Skowron A , Rauszer C. The discernibility mat rices and functions in information system[C] . Intelligent Decision Support Handbook of Applications and Advances of the Rough Set s Theory. Dordrecht : Kluwer Academic Publishers , 1992 : 3312362. [34 ] Wang J ue , Wang J u. Reduction algorithms based on discernibility mat rix : The ordered att ributes method[J ] . J of Computer Science and Technology , 2001,16 (6) : 4892504.
[35] 赵卫东,戴伟辉. 基于特征矩阵的决策表约简研究[J ] . 系统工程理论与实践. 2003 ,23 (3) :65-69.
[36]韩力群.人工神经网络理论、设计及应用[M].北京:化学工业出版社,2002.
[2]盛昭瀚、柳炳祥.客户流失危机分析的决策树方法[J].管理科学学报,2005,8(2):20-25
[3]贾琳、李明.基于数据挖掘的电信客户流失模型的建立与实现[J].计算机工程与应用,2004,40(4):185-187
[4]HUNGSY , YENDC , WANGHY. Applying data mining to telecom churn management[J].Expert Systems with Applications,2006,31(3):515-524.
[5]Mozer Mc. Predicting subscriber Dissatisfaction and Improving Retention in the Wireless Telecommunication Industry[J].IEEE Trans on Neural Networks, 2000,11(3)
[6]Lian Yan. Predicting customer behavior in telecommunications [J].Intelligent Systems,IEEE,2004(2)
[7]Masand B,Datta P,Mani DR.CHAMP:A Prototype for Automated Celluar Churn Prediction[J].Data Mining and Knowledge Discovery,1999(3)
[8]王平.利用数据挖掘实现电信业的客户流失预测分析[D].西南交通大学硕士论文, 2003 .
[9]郭明,郑惠莉,卢毓伟.基于贝叶斯网络的客户流失分析[J].南京邮电大学学报:自然科学版,2005,25(5):79-83
[10]陈文伟,黄金才.数据仓库与数据挖掘[M].北京:人民邮电出版社,2004,5-7
[11] R.Agrawal and R.Srikant, "Fast Algorithms for Mining Association Rules", In Proc. of the 20th Int'1 Conference on Very Large Databases. Santiago, Chile, 1994.
[12] 史忠植.知识发现[M].清华大学出版社,2002.
[13] 陈京民.数据仓库与数据挖掘技术[M].电子工业出版社,2002,1:25-26
[14] CRI SP-DM1.0数据挖掘方法论指南,第1版.CRISP-DM协会.2000.
[15] 耿晓中,张冬梅.数据挖掘综述[J].长春师范学院学报(自然科学版),2006,25 (3):24-27
[16] H.Mannila , H.Toivonen and LVerkamo.Efficient Algorithms for Discovering Association Rules.In AAAI Workshop, Knowledge Discovery in Databases, 181-192, 1994.
[17]Pawlak Z. Rough Set Theoretical Aspects of Reasoning about Data[M].Dordrecht : Kluwer Academic Publishers ,1991.
[18] 龙志勇.数据挖掘在电信行业客户关系管理中的应用[J].信系网络,2003 (12):24-26
[19] Pawlak Z. Rough Sets[J] . International Journal of Computer and Information Science ,1982 , (5) :341 356
[20]石红,沈毅,刘志言等.关于粗糙集理论及应用问题的研究[J].计算机工程,2003,29(3):1-4
[21] Pawlak Z et al.Rough sets[J].Communications of ACM,1995,38(11):89-95.
[22] 刘清.Rough集及Rough推理[M]. 北京: 科学出版社, 2001.
[23] 王国胤.Rough集理论与知识获取[M].西安: 西安交通大学出版社,2001.
[24] 张文修,吴伟志等.粗糙集理论与方法[M].北京: 科学出版社,2001.
[25] 常梨云,王国胤,吴渝. 一种基于Rough Set 理论的属性约简及规则提取方法[J] . 软件学报,1999 ,10 (11) :1206-1211.
[26]芦晓红,陈世权,吴今培.基于可辨识矩阵的启发式属性约简算法及其应用[J].计算机工程,2003,29(1) :56-59.
[27]Hu X H .Knowledge discovery in databases: An attribute-oriented rough s et approach. Canada:University of Regina Pr.,1995.
[28]Hu X H ,CerconeN .Learning in relational databases:A rough set approach. In t.J.of Computational Intelligence,1995,11(2):323-338.
[29]Jakub W. Genetic algorithms in decomposition and classification problems. In:L.Polkows,A.Skowron(eds.):Rough Sets in Knowledge Discovery.Physica-Verlag,Heidelberg,1998(2):471-487
[30] Kryszkiewicz M, Rybinski H. Finding reducts in composed information systems. Int Work shop on Rough Sets and Knowledge Discovery RSKD'93,London:SpringerVerlag,1993:259-268.
[31] Starzyk J, Nelson D E, Sturtz K. Reducts in composed information systems. Bulletin of International Rough Set Society,1999,3(1-2):19-22.
[32] Starzyk J, Nelson D E, Sturtz K. Reducts. A mathematical foundation for improved reduct generation in informations ystems. Journal of Knowledge and Information Systems,2000,2(2):131-146 [33 ] Skowron A , Rauszer C. The discernibility mat rices and functions in information system[C] . Intelligent Decision Support Handbook of Applications and Advances of the Rough Set s Theory. Dordrecht : Kluwer Academic Publishers , 1992 : 3312362. [34 ] Wang J ue , Wang J u. Reduction algorithms based on discernibility mat rix : The ordered att ributes method[J ] . J of Computer Science and Technology , 2001,16 (6) : 4892504.
[35] 赵卫东,戴伟辉. 基于特征矩阵的决策表约简研究[J ] . 系统工程理论与实践. 2003 ,23 (3) :65-69.
[36]韩力群.人工神经网络理论、设计及应用[M].北京:化学工业出版社,2002.