近年來，顧客關係管理成了相當熱門的研究領域，如何提高顧客的忠誠度，並為高貢獻度的客戶提供更好的服務，不但能降低管銷成本，進而獲得實質利益成長，而想達成這些目標，就必需從了解客戶開始。而如何找出客戶相關的行為模式和相關的知識，就成了重要的關鍵。

本研究提出一個以矩陣分群法為基礎分析顧客行為模式的方法。在原先的矩陣分群法中，藉由在顧客與商品所形成的二元行列矩陣中將行與列不斷置換並萃取出一個代表擁有相似行為模式顧客群的密實部份矩陣（Dense sub-matrix），但是這個方法目前在大規模的行列矩陣上運算效率卻還有改進的空間，此外，搜尋出的顧客群組在行為模式的解釋上也有待改進。而本研究所提的方法是將矩陣分群法跟遺傳演算法結合，希望藉由遺傳演算法大量運算與最適化的特性而能在大規模的稀疏行列矩陣上大幅度的縮短運算時間，並且使得所產生的成果具有較佳的解釋品質，以協助管理人員了解不同群組客戶的需求，制定出更佳的行銷決策，提升行銷活動的效益與有效性。

透過實証研究的結果得知，在經過適合度評估方式的最佳化選擇後，遺傳演算法在矩陣分群法的應用上，能獲得運算時間與分析能力顯著的提升，並以實際客戶資料進行行為模式分析找出三個意義顯著的顧客群組，除了可進行基本的推廌應用外，還可對群組深入分析特性，達成類似行銷上購物籃分析的應用。

In recent year, Customer Relationship Management has become a very popular field of study. How to raise customers’ loyalty and provide better services for high contribution customers can not only lower management and sale cost, but also increase revenue growth. To achieve these goals, it’s important to understand customers. Therefore, how to find out behavior model and knowledge related to customers becomes a key point.

In this research, we have proposed a customer behavior analysis method based on Matrix Clustering. In the original Matrix Clustering, the dense sub-matrix was extracted by replacing rows and columns in the binary matrix formed by customers and products constantly. This part of matrix represents a behavior model of a group of similar customers. But the operation efficiency on a large scale of matrix still can be improved; moreover, part of customers’ behavior model will have problems in explanation.

The method proposed in this research is to combine Matrix Clustering with Genetic Algorithms. It is expected that by the ability of large operation and optimization of Genetic Algorithms, operation time on a large scale of sparse matrix should be shortened dramatically and make the result has higher explanation quality to assist management in understanding the demands of different groups of customers. Therefore, management can make better marketing decisions and improve the efficiency of marketing activities.

According to our experiments, it’s been verified that the application of Genetic Algorithms on Matrix Clustering can get significant improvement on both the calculating time and the ability of analysis after the optimal selection of fitness function. And by applying the method on real world customer data, we had discovered three meaningful behaviors for recommending customers and character analysis of both customers and products.

Agrawal, R. and Srikant, R. (1994). “Fast Algorithms for Mining Association Rules”, Proc. 20th VLDB Conf. pp. 487-499.
Berson, A., Smith, S. and Thearing K. (2000). Building Data Mining Application for CRM, McGraw Hill Enterprise Inc.
Billsus, D. and Pazzani, M. J. (1998). “Learning Collaborative Information Filters”, In Proceedings of ICML '98. pp. 46-53.
Conlon, G. (1999). “No Turing Back & Marketing Management” , pp. 50-55.
Davisd, M. (1999). “How to Avoid the 10 Biggest Mistake in CRM” , Journal of Business Strategy, Vol.20, No.6, pp. 22-26.
Don, P., Rogers, M. and Dorf, B. (1999). “Is Your Company Ready for One-to-One Marketing”, Harvard Business Review, pp. 151-160.
Jain, A.K., Murty M.N. and Flynn P.J. (1999). ” Data Clustering: A Review”, ACM Computing Surveys(CSUR), Vol. 31, No. 3, pp. 274-296.
Jiang, J. H., Wang J. H., Chu X. and Yu R. Q. (1997). “Clustering Data Using a Modified Integer Genetic Algorithm,” Analytica Chimica Acta, 354, pp. 263-274.
Kalakota, R. and Robinson, M. (1999). e-Business-Roadmap for success, Addison-Wesley Longman, Inc..
Krishna, K. and Murty, M. N. (1999). “Genetic K-Means Algorithm,” IEEE Transactions on Systems, Man, and Cybernetic-Part B: Cybernetics, 29(3), pp. 433-439.
Man, K.F., Tang, K.S. and Kwong, S. (1999). Genetic Algorithms Concepts and Designs, Spring-Verlag London Limited, pp. 11-17.
Maulik, U. and Bandyopadhyay, S. (2000). “Genetic Algorithm Based Clustering Technique”,Pattern Recognition, 33, pp. 1455-1465.
Michael, J. A. B. and Linoff, G. (2000). Data mining techniques : for marketing, sales, and customer support, John Wiley & Sons, New York.
Mitsuo, G. and Cheng, R. (1997). Genetic Algorithms and Engineering Desigs, A Wiley-Interscience, New York.
Ott, J. (2000), ”Successfully development and Implementing Continuous relationship management”, eBusiness executive report, pp. 26-30
Oyanagi, S., Kubota, K. and Nakase, A. (2001). “Application of Matrix Clustering to Web Log Analysis and Access Prediction.”, WEBKDD2001.
Oyanagi, S., Kubota, K. and Nakase, A. (2001). “Matrix Clustering: a New Data Mining Algorithm for CRM”, IPSJ Journal(in Japaness) Vol.42 No.8., pp. 2156-2166.
Resnick, P., Varian, H. R. and Guest Editors (1997). “Recommender System”, Communication of the ACM, pp. 175-186.
Sarwar, B., Karypis, G., Konstan, J. and Riedl, J. (2000). “Analysis of recommender algorithms for e-commerce”, In Proceedings of the 2nd ACM E-Commerce Conference (EC’00) Minneapolis, pp. 160-162.
Schafer, J. B., Konstan, J. and Riedl, J. (1999). “Recommender System in E-commerce”, Communication of the ACM, pp. 158-159.
Selim, S. Z. and Alsultan, K. (1991). “A Simulated Annealing Algorithm for The Clustering Problem”, Pattern Recognition, 24(10), pp.1003-1008.
Srivastava, J., Wang J. H., Lim E. P. and Hwang S. Y. (2002). “A Case for Analytical Customer Relationship Management”, PAKDD 2002, LNAI 2336, pp.14-27.
Tseng, L. Y. and Yang, S. B. (2000). “A Genetic Clustering Algorithm for Data with Non-Spherical-Shape Clusters,” Pattern Recognition, 33, pp. 1251-1259.
Wayland, R. E. and Cole, P. M. (1997). “Customer Connections: New Strategies for Growth”, Harvard Business School Press.
Witten, I. H. and Frank, E. (1999). Data Mining: practical machine learning tools and techniques with Java implementations, Morgan Kaufmann Publishers.