一种K-MEANS和SOM结合算法在电信客户细分中的应用
摘要
3G时代的到来使得中国电信行业的竞争格局发生了很大的变化,在新的经营环境下,运营商该如何调整各自的经营战略,从而让自己的市场占有率得到有效的提高。此时,能否合理的实施客户细分并准确定位客户的市场需求就成为了关键。数据挖掘技术的研究和发展为电信运营商的客户细分提供了有效的方法。
K-means算法是数据挖掘中发展较早,应用较为成熟的聚类算法之一,其可以用来发现数据集的分布模式,是数据挖掘领域中一种有效的聚类算法。K-means算法也有明显的不足之处,如在聚类之前必须给定最优K值,初始聚类中心的随机确定使得聚类的结果往往具有较大的不确定性等。
本文提出了一种基于距离代价函数的S-K算法,即利用距离代价函数事先确定K-means算法最优K值,弥补了K值必需事先给定的不足;运用SOM神经网络得到初始聚类中心,从而弥补了K-means算法因初始聚类中心随机确定而导致的聚类结果具有很大不确定性的不足,并运用了电信RFM模型来对电信客户群进行更加有效的细分。通过对Iris标准数据集的测试,证明了该算法的有效性。文章的重点是将基于距离代价函数的S-K算法运用在国内某通信运营商的客户细分研究上,得到了较为满意的聚类结果,并以此为运营商提出了具体的具有针对性的营销策略和建议。
With the advent of the era of 3G, the competitive situation of the telecommunication industry has drastically changed in China. How should the operators make use of this golden opportunity to make their market share increase effectively and efficiently? Up to now, the question whether the telecom company can conduct rational customer segmentation and accurately target customer demand has become very critical. Fortunately, research and development of Data mining technology can be adopted as an effective approach to the aforementioned question.
K-means algorithm is one of the earlier clustering algorithms and has relatively more mature applications in data mining; it can be used to identify the distribution pattern of the data set and therefore has been recognized as an effective clustering algorithm in data mining. Of course it should be acknowledged that K-means algorithm has drawbacks, for examples, the optimal K value should be given before clustering, the random determination of the initial cluster center makes the results uncertain, etc.
This paper proposed a S-K algorithm based on the distance cost function, i.e., using a distance cost function in the K-means algorithm for the optimal K value before conducting the clustering; the initial cluster center is obtained using SOM networks rather than being randomly. In the meanwhile, the telecom RFM model is employed to segment telecom customers much more effectively. In sequel, the validity of the proposed algorithm is confirmed through the test of the Iris standard data set. The highlight of the article is the application of S-K algorithm, which is based on the distance cost function, to some telecom company to implement customer segmentation; satisfactory results are derived, which thereby provides more targeted marketing strategies and implications for domestic telecom operation.
K-means算法是数据挖掘中发展较早,应用较为成熟的聚类算法之一,其可以用来发现数据集的分布模式,是数据挖掘领域中一种有效的聚类算法。K-means算法也有明显的不足之处,如在聚类之前必须给定最优K值,初始聚类中心的随机确定使得聚类的结果往往具有较大的不确定性等。
本文提出了一种基于距离代价函数的S-K算法,即利用距离代价函数事先确定K-means算法最优K值,弥补了K值必需事先给定的不足;运用SOM神经网络得到初始聚类中心,从而弥补了K-means算法因初始聚类中心随机确定而导致的聚类结果具有很大不确定性的不足,并运用了电信RFM模型来对电信客户群进行更加有效的细分。通过对Iris标准数据集的测试,证明了该算法的有效性。文章的重点是将基于距离代价函数的S-K算法运用在国内某通信运营商的客户细分研究上,得到了较为满意的聚类结果,并以此为运营商提出了具体的具有针对性的营销策略和建议。
With the advent of the era of 3G, the competitive situation of the telecommunication industry has drastically changed in China. How should the operators make use of this golden opportunity to make their market share increase effectively and efficiently? Up to now, the question whether the telecom company can conduct rational customer segmentation and accurately target customer demand has become very critical. Fortunately, research and development of Data mining technology can be adopted as an effective approach to the aforementioned question.
K-means algorithm is one of the earlier clustering algorithms and has relatively more mature applications in data mining; it can be used to identify the distribution pattern of the data set and therefore has been recognized as an effective clustering algorithm in data mining. Of course it should be acknowledged that K-means algorithm has drawbacks, for examples, the optimal K value should be given before clustering, the random determination of the initial cluster center makes the results uncertain, etc.
This paper proposed a S-K algorithm based on the distance cost function, i.e., using a distance cost function in the K-means algorithm for the optimal K value before conducting the clustering; the initial cluster center is obtained using SOM networks rather than being randomly. In the meanwhile, the telecom RFM model is employed to segment telecom customers much more effectively. In sequel, the validity of the proposed algorithm is confirmed through the test of the Iris standard data set. The highlight of the article is the application of S-K algorithm, which is based on the distance cost function, to some telecom company to implement customer segmentation; satisfactory results are derived, which thereby provides more targeted marketing strategies and implications for domestic telecom operation.
引文
陈京明.数据仓库与数据挖掘技术理论[M].北京:电子工业出版社. 2003.
陈静宇.价值细分-价值细分的驱动模型[J].中国流通经济. 2003.6:53-56.
何敏,张洪伟,张波.模糊ISODATA及在CRM中的应用[J].计算机应用,2005(25):1455-1457
何荣勤. CRM原理?设计?实践[M].北京:电子工业出版社. 2003.
Jiawei Han, Micheline Kamber主编.数据挖掘概念与技术[M],范明,孟小峰,译.北京:机械工业出版社. 2001.
康晓尔.基于数据仓库的数据挖掘技术[M].北京:机械工业出版社. 2004.
李桂华.产业市场营销的市场细分方法与模型研究[J].现代财经,2001.4:55-58.
林盛,肖旭基于RFM的电信客户市场细分方法[J].哈尔滨工业大学学报.2006(5):758-760
王君,樊治平.客户关系管理中客户知识发现的一种分析方法[J].系统工程理论方法应用,2004.1(13):58-62.
王尔平,崔宝灵.数据挖掘技术在电信CRM的应用[J]管理科学,2003.116(5):54-58.
魏敏,李江,万映红.基于目的-价值的客户需求建模研究与实证[J].清华大学学报(自然科学版),2006.S1(46):47- 48.
武森,高学尔,巴斯蒂安等.数据仓库与数据挖掘[M].北京:冶金工业出版社. 2003.
肖胜.国际电信运营商电信市场细分的比较分析[J].世界电信,2001.10(14): 44-47.
谢家平,孔令丞,陈荣秋等.基于QFD的CS指标决策优化模型[J].中国管理科学,2004.6(12):34-41.
严寒冰. CRM推进电信营销模式变革[J].中国电信业,2006.5:48-50.
杨路明,巫宁.客户关系管理理论与实务[M].北京:电子工业出版社. 2004.
杨善林,李永森. K-means算法中的k值优化问题研究[J].系统工程理论与实践,2006.2:97-101.
于珊.数据仓库技术和联机分析处理[M].北京:科学出版社. 1995.
叶孝明,黄祖庆.基于数据挖掘的零售业客户细分研究[J].现代管理科学.2006, 06.
曾勇,黄建华.从英国电信成功转型看我国电信企业深层次转型的三大问题[J].世界电信,2006.7(19):3-4, 34.
张喆，?鹑唬菩≡?数据挖掘技术在客户获取策略中的应用[J].东北大学学报(自然科学版),2003.24(11):1112-1115.
Abraham A, Ramos V. Evolving a Stigmergic Self-Organized Data-Mining[C].The Fourth International Conference on Intelligent Systems.Design and Applications.Hungary:2004
Albrecht Sollner, Mario Rese. Market segmentation and the structure of competition applicability of the strategic group concept for an improved segmentation on industrial markets [J]. Journalof Business Research,2001:25-36.
Ayad H, Kamel M. Topic Discovery from Text Using Aggregation of Different Clustering Methods[A].Cohen R,et al. Advances in Artificial Intelligence:The 15th Conference of the Canadian Society for Computational Studies of Intelligence[C].Calgary,2002. 161-175.
Clausi D A. K-means Iterative Fisher ( KIF) unsupervised clustering algorithm applied to image texture segmentation[J ] . Pattern Recognition , 2002 , 35 :1959 - 1972.
Daniel Barbara, Julia Couto, Yi Li. COOLCAT: An entropy-based algorithm for categorical clustering[C].In: proceedings of the eleventh international conference on Information and Knowledge management, 2002:582-589.
Hong Yao, Howard J. Hamilton..Data & Knowledge Engineering: principles and methods[J], Data & Knowledge Engineering, Elsevier. 2006(2):603-626.
Hruschka, Harald, Fettes, Werner, Probst, Markus.Market segmentation by maximum likelihood clustering using choice elasticities[J].European Journal of Operational Research Volume,2004(3):779-786.
Huang Zhe-xue..Extensions to the k-means algorithms for clustering large data sets with categorical values[J].Data Mining and Knowledge Discovery,1998(3):283-304.
Jainak,Murtymn, Flynn PJ. Data clustering: A review [J].ACM Computer Survey, 1999 (3): 264 - 323.
Jiuh-Biing Sheu. A hybrid fuzzy-optimization approach to customer grouping-based logistics distribution operations [J].Applied Mathematical Modelling,2007 (31): 1048-1066.
Kanungo T,Mount DM,Netanyahu NS. An Efficient K-means ClusteringAlgorithm:Analysis and Implementation[J],IEEE Transactionson Pattern Analysis and Machine Intelligence, 2002,24(7):881-892.
Kohonen T. Self-organized formation of topologically correct feature maps [J], Biological Cybernetics, 1982, 43(1) : 59- 69.
L.A .Zadeh.Fuzzy sets.Information and Control, 1965(8): 338-353.
Smith, Wendell R..Product Differentiation and Market Segmentation as Alt ernative Marketing Strategies[J].Journal of Marketing,1956 :3-8.
Storbacka, Kaj.Segmentation Based on Customer Profitability-Retrospective Analysis of Retail Bank Customer B aces [J].Journal of Marketing Management, 1997 (13):479-492.
Yen-Liang,Chen.Mi-Hao,Kuo,Shin-YiWu,Kei,TangDiscovering recency, frequency, and monetary (RFM) sequential patterns from customers' purchasing data.Electronic Commerce Research and Applications[J] ,2009 (8):241-251
陈静宇.价值细分-价值细分的驱动模型[J].中国流通经济. 2003.6:53-56.
何敏,张洪伟,张波.模糊ISODATA及在CRM中的应用[J].计算机应用,2005(25):1455-1457
何荣勤. CRM原理?设计?实践[M].北京:电子工业出版社. 2003.
Jiawei Han, Micheline Kamber主编.数据挖掘概念与技术[M],范明,孟小峰,译.北京:机械工业出版社. 2001.
康晓尔.基于数据仓库的数据挖掘技术[M].北京:机械工业出版社. 2004.
李桂华.产业市场营销的市场细分方法与模型研究[J].现代财经,2001.4:55-58.
林盛,肖旭基于RFM的电信客户市场细分方法[J].哈尔滨工业大学学报.2006(5):758-760
王君,樊治平.客户关系管理中客户知识发现的一种分析方法[J].系统工程理论方法应用,2004.1(13):58-62.
王尔平,崔宝灵.数据挖掘技术在电信CRM的应用[J]管理科学,2003.116(5):54-58.
魏敏,李江,万映红.基于目的-价值的客户需求建模研究与实证[J].清华大学学报(自然科学版),2006.S1(46):47- 48.
武森,高学尔,巴斯蒂安等.数据仓库与数据挖掘[M].北京:冶金工业出版社. 2003.
肖胜.国际电信运营商电信市场细分的比较分析[J].世界电信,2001.10(14): 44-47.
谢家平,孔令丞,陈荣秋等.基于QFD的CS指标决策优化模型[J].中国管理科学,2004.6(12):34-41.
严寒冰. CRM推进电信营销模式变革[J].中国电信业,2006.5:48-50.
杨路明,巫宁.客户关系管理理论与实务[M].北京:电子工业出版社. 2004.
杨善林,李永森. K-means算法中的k值优化问题研究[J].系统工程理论与实践,2006.2:97-101.
于珊.数据仓库技术和联机分析处理[M].北京:科学出版社. 1995.
叶孝明,黄祖庆.基于数据挖掘的零售业客户细分研究[J].现代管理科学.2006, 06.
曾勇,黄建华.从英国电信成功转型看我国电信企业深层次转型的三大问题[J].世界电信,2006.7(19):3-4, 34.
张喆，?鹑唬菩≡?数据挖掘技术在客户获取策略中的应用[J].东北大学学报(自然科学版),2003.24(11):1112-1115.
Abraham A, Ramos V. Evolving a Stigmergic Self-Organized Data-Mining[C].The Fourth International Conference on Intelligent Systems.Design and Applications.Hungary:2004
Albrecht Sollner, Mario Rese. Market segmentation and the structure of competition applicability of the strategic group concept for an improved segmentation on industrial markets [J]. Journalof Business Research,2001:25-36.
Ayad H, Kamel M. Topic Discovery from Text Using Aggregation of Different Clustering Methods[A].Cohen R,et al. Advances in Artificial Intelligence:The 15th Conference of the Canadian Society for Computational Studies of Intelligence[C].Calgary,2002. 161-175.
Clausi D A. K-means Iterative Fisher ( KIF) unsupervised clustering algorithm applied to image texture segmentation[J ] . Pattern Recognition , 2002 , 35 :1959 - 1972.
Daniel Barbara, Julia Couto, Yi Li. COOLCAT: An entropy-based algorithm for categorical clustering[C].In: proceedings of the eleventh international conference on Information and Knowledge management, 2002:582-589.
Hong Yao, Howard J. Hamilton..Data & Knowledge Engineering: principles and methods[J], Data & Knowledge Engineering, Elsevier. 2006(2):603-626.
Hruschka, Harald, Fettes, Werner, Probst, Markus.Market segmentation by maximum likelihood clustering using choice elasticities[J].European Journal of Operational Research Volume,2004(3):779-786.
Huang Zhe-xue..Extensions to the k-means algorithms for clustering large data sets with categorical values[J].Data Mining and Knowledge Discovery,1998(3):283-304.
Jainak,Murtymn, Flynn PJ. Data clustering: A review [J].ACM Computer Survey, 1999 (3): 264 - 323.
Jiuh-Biing Sheu. A hybrid fuzzy-optimization approach to customer grouping-based logistics distribution operations [J].Applied Mathematical Modelling,2007 (31): 1048-1066.
Kanungo T,Mount DM,Netanyahu NS. An Efficient K-means ClusteringAlgorithm:Analysis and Implementation[J],IEEE Transactionson Pattern Analysis and Machine Intelligence, 2002,24(7):881-892.
Kohonen T. Self-organized formation of topologically correct feature maps [J], Biological Cybernetics, 1982, 43(1) : 59- 69.
L.A .Zadeh.Fuzzy sets.Information and Control, 1965(8): 338-353.
Smith, Wendell R..Product Differentiation and Market Segmentation as Alt ernative Marketing Strategies[J].Journal of Marketing,1956 :3-8.
Storbacka, Kaj.Segmentation Based on Customer Profitability-Retrospective Analysis of Retail Bank Customer B aces [J].Journal of Marketing Management, 1997 (13):479-492.
Yen-Liang,Chen.Mi-Hao,Kuo,Shin-YiWu,Kei,TangDiscovering recency, frequency, and monetary (RFM) sequential patterns from customers' purchasing data.Electronic Commerce Research and Applications[J] ,2009 (8):241-251
© 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 |