臺灣博碩士論文加值系統

行銷模式從傳統的大量廣播式行銷轉變成以顧客為導向，傳統的大量廣播式行銷模式不但增加行銷成本，且利潤回收往往無法達到企業競爭的要求。因此，企業必須主動地去了解顧客的需要並設法滿足。近年來，企業利用資料探勘技術，協助從龐大的企業資料庫中，挖掘出有利於企業競爭的知識。自組映射圖(Self-Organizing Map)，是廣被採用的群聚演算法，它能將多維度的資料投射至二維的平面。然而，傳統自組映射圖無法有效處理種類型資料。我們的主要目的在於提出利用視覺化的群集技術，提供分析人員一交互式的市場區隔及區塊特徵擷取的探勘架構。另一研究目的為，針對自組映射圖於種類型資料上的處理做進一步探討。本研究改良了自組映射圖未能處理種類型資料的缺點，並利用其視覺化特性，使分析人員透過視覺化介面了解資料特徵，進而交互式地決定分群，再搭配屬性導向歸納法擷取隱藏的特徵，最後提供一統計方式檢定特徵間是否有差異。我們透過雛形系統與實際資料驗證，改良式自組映射圖的穩定性及探勘架構的可行性。

Marketing practices have shifted to customer-oriented from traditional mass broadcasting. Traditional mass-broadcasting marketing not only increases marketing cost, but also the profit can’t reach the demand of enterprise’s competition. Therefore, the enterprise must aggressively find customers’ demands and try to satisfy those de-mands. In the past few years, many enterprises have used data mining technology to mine knowledge from huge databases that is useful for competition. SOM (Self-Organizing Maps) is a well-known clustering algorithm which can project multi-dimensional data onto a two-dimensional plane. However, the traditional SOM doesn’t work for categorical values. This thesis proposes an interactive mining model that uses a visual clustering technique and applies it to market segmentation and seg-ments’ feature extraction. Besides, we also investigate how to improve the SOM algo-rithm such that it can be used on categorical data. Users can utilize the visual character-istic of the SOM to analyze the features of data on the visualized map and interactively perform data clustering. And then users can use the attribute-oriented induction algo-rithm to retrieve the hidden patterns in individual clusters, and at last we provide a sta-tistics method to verify the differences among patterns. We develop a prototype and test on synthetic data and real-world data to verify the stability of the improved SOM and the practicability of the proposed data-mining architecture.

中文摘要 i
ABSTRACT ii
誌謝 iii
目錄 iv
表目錄 vi
圖目錄 vii
壹、緒論 1
1.1 研究背景及動機 1
1.2 研究目的 2
1.3 研究範圍及限制 2
1.4 論文架構 3
貳、文獻探討 4
2.1 市場區隔 4
2.2 資料庫知識探索與資料探勘 4
2.3 自組映射圖介紹 5
2.4 概念階層 7
參、研究架構 9
3.1 市場區隔及區塊特徵擷取探勘架構 9
3.2 探勘前準備 11
3.2.1 前置處理 11
3.2.2 概念階層建置 11
3.3 改良式自組映射圖 13
3.3.1 種類型資料的距離問題 13
3.3.2 種類型資料於自組映射圖上的調整方式 14
3.3.3 參數及初始值設定介紹 17
3.4 特徵分析 18
3.4.1 特徵擷取 18
3.4.2 特徵關聯 19
3.4.3 特徵比對 19
肆、實驗與雛型展示 20
4.1 改良式自組映射圖驗證實驗 20
4.1.1 神經元數目對資料分群上的影響 20
4.1.2 調整半徑對分群品質的影響 28
4.1.3 不同自組映射圖初始值對分群品質之影響 30
4.1.4 資料輸入順序對分群品質之影響 31
4.1.5 傳統與改良式自組映射圖於種類型資料之比較 32
4.1.6 神經元數與資料筆數於所需時間的實驗 37
4.1.7 混合類型資料實驗 40
4.1.8 改良式自組映射圖驗證實驗結論 42
4.2 資料分群及群組特徵擷取之應用 43
4.2.1 賣場會員消費特徵探勘 43
4.2.2 UCI電子資料庫成人收入調查資料 47
伍、結論及未來研究 51
5.1 結論 51
5.2 未來研究 51
參考文獻 53
附錄 56

1. 方世榮譯，Philip Kotler著，1996，”行銷管理學－分析、計劃、執行、與控制”，2版，臺北：東華書局，頁282-304。
2. 林彥成、蘇育葳、林旻宏、許中川，2002，”應用資料探勘技術於市場區隔及區塊特徵擷取”，第一屆服務業行銷暨管理學術研討會。
3. 林彥成、林旻宏、蘇育葳、許中川，2002，”改良式自組映射圖應用於市場區隔及區塊特徵擷取”， 第十三屆國際資訊管理學術研討會。
4. 苑守慈、張瑋倫，1999，”資料挖掘與顧客保留/顧客利潤創造”，民國八十八年全國計算機會議論文集。
5. 許中川、洪鋕鋒，1997，”資料庫知識發掘前置處理與欄位拆解”，第三屆國際資訊管理研究暨實務研討會論文集，頁362-369。
6. 許中川、洪鋕鋒，1998，“以屬性階層支援資料分析”，第九屆國際資訊管理學術研討會論文集。
7. Ahlberg, C., C. Williamson and B. Sheniderman , 1992a, “Dynamic Queries for Information Exploration：An Implementation and Evaluation”, Proc. CHI Conf., ACM Press, pp.619-626.
8. Ahlberg, C., Dec.1996, “Spotfire: an information exploration environment”, ACM SIGMOD, Vol.25, No.4, pp.25-29.
9. Chen, M. S., J. Han and P. S. Yu, Dec. 1996, “Data Mining﹕An Overview from a Database Perspective”, IEEE Transactions on Knowledge and Data Engineering, Vol. 8, No. 6, pp. 866-883.
10. Fayyad, U. and R. Uthurusamy, Nov. 1996, “Data mining and knowledge discov-ery in databases”, Communications of the ACM, Vol. 39, pp. 24 — 26.
11. Gruca, T. S., B. R. Klemz and E. A. F. Petersen, Jun. 1999, “Mining Sales Data Using A Neural Network Model of Market Response”, ACM SIGMOD, Vol. 1, No. 1, pp. 39-43.
12. Han, J., and Y. Fu., July 1994, “Dynamic generatioin and refinement of concept hierarchies for knowledge discovery in databases”, In Proc. AAAI’94 Workshop Knowledge Discovery in Database(KDD’94), pp. 157-168.
13. Han, J., Mar. 1998, “Towards On-Line Analytical Mining in Large Databases”, ACM SIGMOD RECORD, Vol. 27, No. 1, pp. 97-107.
14. Han, J. and M. Kamber, 2001, “Data Mining Concepts and Techniques”, San Fran-cisco: Morgan Kaufmann, pp181-194.
15. Hinneburg, A., D. Keim and M. Wawryniuk, September/October 1999, “HD-Eye: visual Mining of High-Dimensional Data”, IEEE Computer Graphics and Applica-tions, Vol. 19 pp. 22-31.
16. Hsu, C.C. and C.F. Hong, 1998, “Supporting Data Analysis with Attribute Hierar-chy”, Proc. of Int’l Symposium on Intelligent Data Engineering and Learning, pp.201-209
17. Ioannidis, Y., Dec. 1996, “Dynamic Information Visualization”, ACM SIGMOD, Vol. 25, No. 4, pp.16-20.
18. Kohonen T, Oja E and Ruohonen M, 1974, “Adaptation of a linear system to a fi-nite set of patterns occurring in an arbitrarily varying order”. Acta Polytech. Scand. Elect. Eng. 25.
19. Kohonen, T., 1990, “The Self-Organizing Map,” Proceeding of the IEEE, 789, pp1464-1480.
20. Kohonen, T., Samuel Kaski, Krista Lagus, Jarkko Salojarvi, Jukka Honkela, Vesa Paatero, and Antti Saarela, May 2000, “Self organization of a massive document collection”. IEEE Transactions on Neural Networks. Vol. 11, No. 3.
21. Ng, K. S., H. Liu and H. B. Kwah, 1998, “A Data Mining Application﹕Customer Retention at the Port of Singapore Authority PSA”, Proceeding of the 1998 ACM SIGMOD Conference, Vol. 27, pp. 522-525.
22. Ribarsky, W., J. Katz, F. Jiang and A. Holland, September/October 1999, “Discovery Visualization Using Fast Clustering”, IEEE Computer Graphics and Appli-cations, Vol. 19, pp. 32-39.
23. Rohrer, R. M. and J. Sibert, September/October 1999, “A Shape-based Visual In-terface for Text Retrieval”, IEEE Computer Graphics and Applications, Vol. 19, pp. 40-46.
24. Shneiderman, B., Nov. 1994, “Dynamic Queries for visual Information Seeking”, IEEE Software, Vol. 11, No. 6, pp.70-77.
25. Simoudis, E., Oct. 1996, “Reality Check for Data Min-ing”, IEEE EXPERT, pp. 26-33.
26. Srikant, R. and R. Agrawal, Jun. 1996, “Mining Quantitative Association Rules in Large Relational Tables”, Proceedings of the 1996 ACM SIGMOD International Conference on Management of Data, pp. 1-12.
27. Vesanto, J., Esa Alhoniemi, Johan Himberg, Kimmo Kiviluoto, and Jukka Parviai-nen, March 1999, “Self-Organizing Map for Data Mining in Matlab: the SOM Toolbox”, Simulation News Europe, (25):54.
28. Vesanto, J. and Esa Alhoniemi, 2000, “Clustering of the Self-Organizing Map”, IEEE Transactions on Neural Networks, Accepted for publication.