移动通信经营分析系统的构建与客户流失分析
|
摘要
近年来，数据仓库和数据挖掘等新技术的迅速发展为决策支持系统(DSS)的发展开辟了新途径。将决策支持系统由传统的以模型库系统为主体，通过定量分析进行辅助决策转向由数据驱动进行辅助决策，使计算机辅助决策能力上了一个新台阶。目前开发的综合DSS是以数据仓库技术为基础，以联机分析处理和数据挖掘工具为手段进行实施的一整套解决方案。
本文以移动通信经营分析系统为研究背景，根据移动通信行业的数据特点，按照“自底向上”的基本原则，构建面向业务主题的数据集市，并在此基础上最终形成面向整个业务系统的中央数据仓库。在成功构建数据仓库系统之后，针对移动通信行业日益突出的客户流失问题，本文采用了多种理论相互融合的思想，将神经网络和决策树技术相结合，构建客户流失分析模型。文章对神经网络和决策树技术进行了深入的分析，研究其各自的优缺点，并分析了将这两种技术结合在一起的可能性及优势。在客户流失模型的构造过程中，本文针对神经网络算法的缺陷运用了新的改进算法，提高了训练的精度和收敛速度。同时，在传统的决策树算法的分裂准则中成功引入了误分代价的因素，从而提高了分类模型的准确性和适用性。最后通过实际数据对模型进行了应用评估，结果表明这种基于神经网络和决策树技术的预测模型能够对客户流失情况做出准确的预测，达到了商业使用的要求。
During the past several years, quick development of Data Warehouse and Data Mining has opened a new approach for Decision Support System (DSS). The transformation from decision support system based on quantitative analysis dominated by modeling system to data-driven system has made a new improvement in computer-aided decision ability. At current, generalized DSS is a set of schemes based on Data Warehouse with the tools of On-Line Analytical Processing and Data Mining.
With generalized Business Analysis System of mobile communication as the research background, and according to characters of data in this field, this thesis is based on bottom-up principle to construct business-subject-oriented Data Marts, and ultimately forms a central data warehouse oriented at the whole business system. After successful construction of data warehouse system, this thesis crossly applies several theories to combine technologies of neural network and decision tree. Thus a model of the analysis of customer chum is built to solve the emerging problem of customer churn from mobile communication companies. This thesis also provides an in-depth analysis of neural network and decision tree to find out their respective merits and drawbacks, and performs a research on the superiority of the combination of these two technologies. During the process of constructing the model of analyzing customer churn, an improved algorithm is applied in this thesis directed toward drawbacks of computer network and thus raises the training accuracy and rapidity of convergence. At the same time, after successful absorbent of the factor of misclassification cost into splitting principle of decision tree algorithm, the classification model gets a high improvement in accuracy and adaptation. From the evaluation on models with actual data, it demonstrates that such a predictive model based on neural network and decision tree can provide a comparatively accurate prediction of customer churn and satisfy the requirement of commercial application.
本文以移动通信经营分析系统为研究背景，根据移动通信行业的数据特点，按照“自底向上”的基本原则，构建面向业务主题的数据集市，并在此基础上最终形成面向整个业务系统的中央数据仓库。在成功构建数据仓库系统之后，针对移动通信行业日益突出的客户流失问题，本文采用了多种理论相互融合的思想，将神经网络和决策树技术相结合，构建客户流失分析模型。文章对神经网络和决策树技术进行了深入的分析，研究其各自的优缺点，并分析了将这两种技术结合在一起的可能性及优势。在客户流失模型的构造过程中，本文针对神经网络算法的缺陷运用了新的改进算法，提高了训练的精度和收敛速度。同时，在传统的决策树算法的分裂准则中成功引入了误分代价的因素，从而提高了分类模型的准确性和适用性。最后通过实际数据对模型进行了应用评估，结果表明这种基于神经网络和决策树技术的预测模型能够对客户流失情况做出准确的预测，达到了商业使用的要求。
During the past several years, quick development of Data Warehouse and Data Mining has opened a new approach for Decision Support System (DSS). The transformation from decision support system based on quantitative analysis dominated by modeling system to data-driven system has made a new improvement in computer-aided decision ability. At current, generalized DSS is a set of schemes based on Data Warehouse with the tools of On-Line Analytical Processing and Data Mining.
With generalized Business Analysis System of mobile communication as the research background, and according to characters of data in this field, this thesis is based on bottom-up principle to construct business-subject-oriented Data Marts, and ultimately forms a central data warehouse oriented at the whole business system. After successful construction of data warehouse system, this thesis crossly applies several theories to combine technologies of neural network and decision tree. Thus a model of the analysis of customer chum is built to solve the emerging problem of customer churn from mobile communication companies. This thesis also provides an in-depth analysis of neural network and decision tree to find out their respective merits and drawbacks, and performs a research on the superiority of the combination of these two technologies. During the process of constructing the model of analyzing customer churn, an improved algorithm is applied in this thesis directed toward drawbacks of computer network and thus raises the training accuracy and rapidity of convergence. At the same time, after successful absorbent of the factor of misclassification cost into splitting principle of decision tree algorithm, the classification model gets a high improvement in accuracy and adaptation. From the evaluation on models with actual data, it demonstrates that such a predictive model based on neural network and decision tree can provide a comparatively accurate prediction of customer churn and satisfy the requirement of commercial application.
引文
[1] M.S.Scott Morton, "Management Decision Systems: Computer-Based Support for Decision Making", Division of Research, Graduate School of Business Administration, Harvard University, 1971.
[2] R.H.Sprague and E.D.Calsson, Building Effective Decision Support Systems, Prentice Hall, 1982.
[3] M.S.Sliver, Systems that Support Decision Makers, Wiley and Sons, 1991.
[4] R.H.Bonczek, C.W.Holsapple and A.Whinston, Foundations of Decision Support Systems, Academic Press, 1981.
[5] Sean B Eom, "The Intellectual Development and structure of Decision Support System", Omega, 26(5):639-657, 1998.
[6] Peter G.W.Keen and M,S.Scott Morton, Decision Support Systems: An Organizational Perspective, Reading, MA:Addison-Wesley, Inc., 1978.
[7] W.H.Inmon, Building the Data Warehouse, John Wiley and Sons, 1996.
[8] S.Adlen, "Exploring the Data Warehouses Organizational and Cultural Issues", D.atabase Programming and Design, 8(6):25-27, 1995.
[9] J.Ladley, "Operational Data Stores:Building an Effective Strategy", Data Warehouse: Pratical Advice form the Experts, Prentice Hall, Englewood Cliffs, NJ, 1997.
[10] E.Appleton, "The Right Server for Your Data Warehouse", Datamation March, 56-58,1996.
[11] H.H.Watson and B.J.Haley, "Data Warehousing: A Framework and.Survey of Current Practices", Journal Data Warehousing, 2(1): 10-17, 1997.
[12] S.R.Gardner, "Building the Data Warehouse", Communication of ACM, 41(9):52-60, 1998.
[13] W.H.Inmon著，王志海，林友芳等译，《数据仓库》，北京：机械工业出版社，2000。
[14] A.Sen and V.S.Jacob, "Industrial Strength Data Warehousing: Why Process is so Important and so often Ignored", Communication of ACM, 41(9):29-31, 1998.
[15] 段云峰，吴唯宁等编著，《数据仓库及其在电信领域中的应用》，北京：电子工业出版社，2003。
[16] Jiawei Han,Micheline Kamber著，范明，孟晓峰等译，《数据挖掘：概念与技术》，北京：机械工业出版社，2001。
[17] David Hand，Heikki Mannila等著，张银奎，廖丽等译，《数据挖掘原理》，北京：
机械工业出版社，2003。
[18] R.E.Stepp and R.S.Michalski, "Conceptual Clustering: Inventing Goaloriented Classifications of Structured Objects", Machine Learning' An AI Approach, Morgan Kaufmann Publishers, 2:471-498, 1986.
[19] J.W.Han, "Data Mining Technique", In Proceedings of ACM SIGMOD International Conference on Management of Data, Montreal, Canada, 1996.
[20] H.J.Lu, R.Setiono and H.Liu, "Effective Data mining Using Neural Networks", IEEE Transactions on Knowledge and Data Engineering, 8(6):957-961, 1996.
[21] H.Mannila, H.Toivonen and I.Verkamo, "Efficient Algorithms for Discovering Association Rules", In AAAI Workshop, Knowledge Discovery in Databases, 181-192, 1994.
[22] R.Agrawal and R.Srikant, "Fast Algorithms for Mining Association Rules", In Proc. of the 20th Int'l Conference on Very Large Databases, Santiago, Chile, 1994.
[23] D.Fisher, "Optimization and Simplification of Hierarchical Clusterings", In Proe.of the First Int'l Conference on Knowledge Discovery and Data Mining, 118-123, 1995.
[24] A.Arning, R.Agrawal and P.Raghavan, "A Linear Method for Deviation Detection In Large Databases", In KDD-96 Proceedings, Second International Conference on Knowledge Discovery and Data Mining, 1996.
[25] F.Rosenblatt, "The Pereeptron: A probabilistic model for information storage and organization in the brain", Psychological Review, 65(6):386-408, 1958.
[26] J.J.Hopfield, "Neurons with Graded Response Have Collective Computational Properties Like Those of Two-state Neurons", Proceedings of the National Academy of Science, 81:3088-3092, 1984.
[27] D.E.Rumelhart, G.E.Hinton and R.J.Williarns, "Learning representations by back-propagating errors", Nature, 323(9):533-536, 1986.
[28] D.E.Rumelhart, J.L.Mcclelland and the PDP research group, Parallel Distributed Processing, MIT Press, Cambridge, MA, 1986.
[29] J.R.Quinlan and R.L.Rivest, "Inferring Decision Trees Using the Minimum Description Length Principle", Information and Computation, 80(3):227-248, 1989.
[30] J.Sehlimmer and D.Fisher, "A Case Study of Incremental Concept Induction", In Proceedings of the Fifth National Conference on Artificial Intelligence, Menlo Park,CA:AAAI Press, 496-501, 1986.
[31] EE.Utlgoff, IDS:Ineremental ID3, In Proceedings of ICML-88, San Marco, CA,1998.
[32] J.R .Quinlan, "Induction of Decision Trees", Machine Learning, 1(1):81-106, 1986.
[33] J.R.Quinlan, C4.5:Programs or machine learning, Morgan aufmann Publishers, 1993.
[34] L.Breiman, J.H.Friedman, R.A.Olshen and C.J.Stone, Classification and Regression Trees, Chapman & Hall, New York, 1984.
[35] S.Sarawagi, S.Thomas and R.Agrawal, "Integrating association rule mining with relational database systems: Alternatives and implications", In Proceedings of ACM SIGMOD International Conference, 343-354, 1998.
[36] M.Minsky, "Logical versus analogical or symbolic versus connectionist or neat versus scruffy", AI Magazine, 12(2):34-51, 1991.
[37] R.Sun, L.A.Bookman, "How Do Symbols and Networks Fit Together: A Report from the AAAI Workshop on Integrating Neural and Symbolic Processes", AI Magazine, 14(2), 1993.
[38] CRISP 1.0 Process and User Guide, http://www.crisp-dm.org.
[39] D.E.Rumelhart, G.E.Hinton and R.J.Williams, "Learning Representations by Back-propagating Errors", Nature, 323(9):533-536, 1986.
[40] C.Charalambous, "A Conjugate Gradient Algorithm For the Efficient Training of Artificial Neural Networks" , IEEE Proceedings Part G, 139(3):301-310, 1992.
[41] M.T.Hagan and M.Menhaj, "Training feedforward networks with the marquardt algorithm", IEEE Transactions on Neural Networks, 5(6):989-993, 1994.
[42] T.P.Vogl, J.K.Mangis, A.K.Rigler, W.T.Zink and D.L.Alkon, "Accelerating the convergence of the back-propagation method", Biological Cybernetics, (59):257-263, 1988.
[43] M.J.Pazzani, C.J.Merz, P.Murphy, K.Ali, T.Hume and C.Brunk, "Reducing Misclassification Costs", In Proceedings of the 11th International Conference of Machine Learning, Morgan Kaufmann, 217-225, 1994.
[44] K.M.Ting, "Inducing Cost-sensitive Trees Via Instance Weighting", In Proceedings of The Second European Symposium on Principles of Data Mining and Knowledge Discovery, LNAI-1510, 139-147, 1998.
[45] P.R.Cohen, D.Jensen, "Overfitting Explained", Preliminary Papers of the Sixth International Workshop on Artificial Intelligence and Statistics, 115-122, 1997.
[46] F.Esposito, D.Malerba and G.Semeraro, "A Comparative Analysis of Methods for Pruning Decision Trees", IEEE Transactions on Pattern Analysis and Machine Intelligence, 19(5):476-491, 1997.
[47] J.R.Quinlan, "Simplifying Decision Trees", International Journal of Man-Machine Studies, 27(3):221-234, 1987.
[48] D.Malerba,F.Esposito and G.Semeraro, "A Further Comparison of Simplification Methods for Decision-Tree Induction", Learning From Data:Artificial Intelligence and Statistics V, Berlin:Springer, 112:365-374, 1996.
[2] R.H.Sprague and E.D.Calsson, Building Effective Decision Support Systems, Prentice Hall, 1982.
[3] M.S.Sliver, Systems that Support Decision Makers, Wiley and Sons, 1991.
[4] R.H.Bonczek, C.W.Holsapple and A.Whinston, Foundations of Decision Support Systems, Academic Press, 1981.
[5] Sean B Eom, "The Intellectual Development and structure of Decision Support System", Omega, 26(5):639-657, 1998.
[6] Peter G.W.Keen and M,S.Scott Morton, Decision Support Systems: An Organizational Perspective, Reading, MA:Addison-Wesley, Inc., 1978.
[7] W.H.Inmon, Building the Data Warehouse, John Wiley and Sons, 1996.
[8] S.Adlen, "Exploring the Data Warehouses Organizational and Cultural Issues", D.atabase Programming and Design, 8(6):25-27, 1995.
[9] J.Ladley, "Operational Data Stores:Building an Effective Strategy", Data Warehouse: Pratical Advice form the Experts, Prentice Hall, Englewood Cliffs, NJ, 1997.
[10] E.Appleton, "The Right Server for Your Data Warehouse", Datamation March, 56-58,1996.
[11] H.H.Watson and B.J.Haley, "Data Warehousing: A Framework and.Survey of Current Practices", Journal Data Warehousing, 2(1): 10-17, 1997.
[12] S.R.Gardner, "Building the Data Warehouse", Communication of ACM, 41(9):52-60, 1998.
[13] W.H.Inmon著，王志海，林友芳等译，《数据仓库》，北京：机械工业出版社，2000。
[14] A.Sen and V.S.Jacob, "Industrial Strength Data Warehousing: Why Process is so Important and so often Ignored", Communication of ACM, 41(9):29-31, 1998.
[15] 段云峰，吴唯宁等编著，《数据仓库及其在电信领域中的应用》，北京：电子工业出版社，2003。
[16] Jiawei Han,Micheline Kamber著，范明，孟晓峰等译，《数据挖掘：概念与技术》，北京：机械工业出版社，2001。
[17] David Hand，Heikki Mannila等著，张银奎，廖丽等译，《数据挖掘原理》，北京：
机械工业出版社，2003。
[18] R.E.Stepp and R.S.Michalski, "Conceptual Clustering: Inventing Goaloriented Classifications of Structured Objects", Machine Learning' An AI Approach, Morgan Kaufmann Publishers, 2:471-498, 1986.
[19] J.W.Han, "Data Mining Technique", In Proceedings of ACM SIGMOD International Conference on Management of Data, Montreal, Canada, 1996.
[20] H.J.Lu, R.Setiono and H.Liu, "Effective Data mining Using Neural Networks", IEEE Transactions on Knowledge and Data Engineering, 8(6):957-961, 1996.
[21] H.Mannila, H.Toivonen and I.Verkamo, "Efficient Algorithms for Discovering Association Rules", In AAAI Workshop, Knowledge Discovery in Databases, 181-192, 1994.
[22] R.Agrawal and R.Srikant, "Fast Algorithms for Mining Association Rules", In Proc. of the 20th Int'l Conference on Very Large Databases, Santiago, Chile, 1994.
[23] D.Fisher, "Optimization and Simplification of Hierarchical Clusterings", In Proe.of the First Int'l Conference on Knowledge Discovery and Data Mining, 118-123, 1995.
[24] A.Arning, R.Agrawal and P.Raghavan, "A Linear Method for Deviation Detection In Large Databases", In KDD-96 Proceedings, Second International Conference on Knowledge Discovery and Data Mining, 1996.
[25] F.Rosenblatt, "The Pereeptron: A probabilistic model for information storage and organization in the brain", Psychological Review, 65(6):386-408, 1958.
[26] J.J.Hopfield, "Neurons with Graded Response Have Collective Computational Properties Like Those of Two-state Neurons", Proceedings of the National Academy of Science, 81:3088-3092, 1984.
[27] D.E.Rumelhart, G.E.Hinton and R.J.Williarns, "Learning representations by back-propagating errors", Nature, 323(9):533-536, 1986.
[28] D.E.Rumelhart, J.L.Mcclelland and the PDP research group, Parallel Distributed Processing, MIT Press, Cambridge, MA, 1986.
[29] J.R.Quinlan and R.L.Rivest, "Inferring Decision Trees Using the Minimum Description Length Principle", Information and Computation, 80(3):227-248, 1989.
[30] J.Sehlimmer and D.Fisher, "A Case Study of Incremental Concept Induction", In Proceedings of the Fifth National Conference on Artificial Intelligence, Menlo Park,CA:AAAI Press, 496-501, 1986.
[31] EE.Utlgoff, IDS:Ineremental ID3, In Proceedings of ICML-88, San Marco, CA,1998.
[32] J.R .Quinlan, "Induction of Decision Trees", Machine Learning, 1(1):81-106, 1986.
[33] J.R.Quinlan, C4.5:Programs or machine learning, Morgan aufmann Publishers, 1993.
[34] L.Breiman, J.H.Friedman, R.A.Olshen and C.J.Stone, Classification and Regression Trees, Chapman & Hall, New York, 1984.
[35] S.Sarawagi, S.Thomas and R.Agrawal, "Integrating association rule mining with relational database systems: Alternatives and implications", In Proceedings of ACM SIGMOD International Conference, 343-354, 1998.
[36] M.Minsky, "Logical versus analogical or symbolic versus connectionist or neat versus scruffy", AI Magazine, 12(2):34-51, 1991.
[37] R.Sun, L.A.Bookman, "How Do Symbols and Networks Fit Together: A Report from the AAAI Workshop on Integrating Neural and Symbolic Processes", AI Magazine, 14(2), 1993.
[38] CRISP 1.0 Process and User Guide, http://www.crisp-dm.org.
[39] D.E.Rumelhart, G.E.Hinton and R.J.Williams, "Learning Representations by Back-propagating Errors", Nature, 323(9):533-536, 1986.
[40] C.Charalambous, "A Conjugate Gradient Algorithm For the Efficient Training of Artificial Neural Networks" , IEEE Proceedings Part G, 139(3):301-310, 1992.
[41] M.T.Hagan and M.Menhaj, "Training feedforward networks with the marquardt algorithm", IEEE Transactions on Neural Networks, 5(6):989-993, 1994.
[42] T.P.Vogl, J.K.Mangis, A.K.Rigler, W.T.Zink and D.L.Alkon, "Accelerating the convergence of the back-propagation method", Biological Cybernetics, (59):257-263, 1988.
[43] M.J.Pazzani, C.J.Merz, P.Murphy, K.Ali, T.Hume and C.Brunk, "Reducing Misclassification Costs", In Proceedings of the 11th International Conference of Machine Learning, Morgan Kaufmann, 217-225, 1994.
[44] K.M.Ting, "Inducing Cost-sensitive Trees Via Instance Weighting", In Proceedings of The Second European Symposium on Principles of Data Mining and Knowledge Discovery, LNAI-1510, 139-147, 1998.
[45] P.R.Cohen, D.Jensen, "Overfitting Explained", Preliminary Papers of the Sixth International Workshop on Artificial Intelligence and Statistics, 115-122, 1997.
[46] F.Esposito, D.Malerba and G.Semeraro, "A Comparative Analysis of Methods for Pruning Decision Trees", IEEE Transactions on Pattern Analysis and Machine Intelligence, 19(5):476-491, 1997.
[47] J.R.Quinlan, "Simplifying Decision Trees", International Journal of Man-Machine Studies, 27(3):221-234, 1987.
[48] D.Malerba,F.Esposito and G.Semeraro, "A Further Comparison of Simplification Methods for Decision-Tree Induction", Learning From Data:Artificial Intelligence and Statistics V, Berlin:Springer, 112:365-374, 1996.