資料探勘(Data Mining)技術是由資料中挖掘出有用的特徵樣式程序，進而利用發掘出之樣式來解釋現存的行為或預測未來的結果，這些是一些傳統資料分析工具所無法解決的問題，本文之目標在於應用資料探勘技術從業界交易資料庫中，挖掘感興趣的關聯規則(Association Rule)，進而發現銷售商品彼此之間的關聯度以及商品和目標間的關係。
網際網路的興起對目前的生活型態及商業模式產生關鍵性的影響，而電子商務的交易方式和一般商店的消費方式有什麼不同，關健在於區域，傳統的行銷方式行銷會以商家附近開始進行行銷，但電子商務卻大大的不同，它打破了傳統的購物方式，因為在電子平台的世界中已經沒有所謂的區域，若可分析出各個區域的購物行為，便可得知該區域所欠缺商品，進而可針對不同的區域進行不同類別商品的行銷，並可從分析資料中，輕易的得到潛藏的客戶群。
本文首先會使用Apriori演算法，分析交易資料庫中的資訊，但也由於Apriori 演算法有相當大的瓶頸，所以後續所提出的頻繁型樣樹法(Frequent Pattern Tree) 與頻繁型樣串列法(Frequent Pattern List)皆是屬於可彌捕Apriori的方法，在本文中將深入探討Apriori與FP-tree以及FP-list的建構方式，並以電子商務業界中實際交易的資料庫進行分析比較，可知FP-tree分析方式所帶來的優點。

There is no doubt that the rapid growth of e-commerce has significantly impacted on consumers’ purchase behaviors and an organization’s marketing strategies. One may ask what is the difference between e-commerce websites and traditional brick-and-mortar stores. A major difference is that a traditional store begins its business by attracting customers in the area or plaza where it is located, while an e-commerce website enables the marketers to access customers globally without boundaries. Therefore, by analyzing the consumers’ purchase behaviors collected from different regions, information about what products are in need and what marketing strategies to be applied can be yielded. Also, with such information about market segmentation, a marketer can easily target and reach the potential customers by sending right messages.
Data Mining, a technology that scrutinizes and sorts databases for meaningful correlations or patterns, is used to explain a current phenomenon or to provide models to predict future results. This method is incomparable because it provides information that traditional analyzing tools are not able to generate. The objective of this study is to apply the Association Rule to the marketing database and find out the correlations among the products as well as the relationships between the products and the target segments.
Initially, this study attempts to use Apriori algorithm to analyze the data in the marketing database. However, due to some deficiency of Apriori, Frequent pattern tree and Frequent pattern list, the latest and most efficient algorithms, are later adopted. In this study, the constructions of Apriori, Frequent pattern tree, and Frequent pattern list are thoroughly discussed. Moreover, the author applies these methods to the marketing database. By analyzing and comparing them, this study suggests that the Frequent pattern related methods in Data Mining will be advantageous and profitable tools for all marketers of e-commerce.

目錄
第一章、 緒論........................1
1.1  前言............................1
1.2  研究動機........................3
1.3  區域性購物分析..................3

第二章、 關聯規則探討................6
2.1  資料探勘與關聯規則..............6
2.2  Apriori演算法...................9
2.3  FP-Tree演算法...................13
2.4  FP-list演算法...................19
2.5  調整FP-list演算法之技術研究.....26

第三章、 實驗架構....................36
3.1 實驗說明.........................36
3.2 實驗架構.........................37
3.3 原始資料轉換至分析陣列...........44
3.3.1  傳統掃描方式至分析陣列........44
3.3.2  SQL語法方式至分析陣列.........52
3.4 資料分析架構.....................55
3.4.1  傳統掃描方式架構Apriori演算法.55
3.4.2  SQL語法方式架構Apriori演算法..62
3.4.3  綜合FP-Tree演算法以及FP-List演算法架構方式.65

第四章、 數據分析與探討..............70
4.1  傳統掃描與SQL分析...............70
4.2  Apriori演算法與FP-Tree分析......74
4.3  區域性分析所帶來的好處並綜合分析與探討.76

第五章、 結論........................79
5.1 結論.............................79
5.2 未來研究方向.....................80

附錄一、 參考文獻....................82
附錄二、 程式執行畫面................84

圖目錄
圖2.1  FB-Tree結構流程圖 1.............15
圖2.2  FB-Tree結構流程圖 2.............16
圖2.3  FB-Tree結構流程圖 3.............16
圖2.4  改良之FB-list結構流程圖 Node C..29
圖2.5  改良之FB-list結構流程圖 Node D..29
圖2.6  改良之FB-list結構流程圖 Node B..30
圖2.7  改良之FB-list結構流程圖 Node A..30
圖2.8  改良之FB-list結構流程圖 Node F..31
圖2.9  改良之FB-list結構流程圖 Node E..31
圖3.1  程式設計之流程..................39
圖3.2  傳統掃描方式至分析陣列流程......45
圖3.3  SQL語法方式至分析陣列流程.......52
圖3.4  傳統掃描方式架構Apriori演算法流程.56
圖3.5  轉至Apriori Table流程...........61
圖3.6  SQL語法方式架構Apriori演算法流程..62
圖3.7  FP-Tree演算法架構流程...........66
圖4.1  資料轉換至分析陣列結果分析......71
圖4.2  Apriori演算法進行資料分析結果...72
圖4.3  Apriori演算法與FP-Tree分析......75
圖4.4  交易資料庫分佈圖................77
圖4.5  最小支持度設定與執行效能分佈圖..78

表目錄
表 2.1  範例交易資料庫................9
表 2.2  頻繁商品項目..................11
表 2.3  頻繁商品項目集合..............11
表 2.4  組合頻繁商品項目..............12
表 2.5  Apriori最後頻繁項目的輸出.....13
表 2.6  FB-Tree最後頻繁項目輸出.......17
表 2.7  FP-Tree輸出結果...............18
表 2.8  交易記錄轉為Bit輸出...........20
表 2.9  FP-List節點C..................20
表 2.10  Bit Counting節點.............21
表 2.11  FP-List節點D.................21
表 2.12  Bit Counting節點D............22
表 2.13  FP-List節點B.................22
表 2.14  Bit Counting節點B............23
表 2.15  FP-List節點A.................23
表 2.16  Bit Counting節點A............24
表 2.17  FP-List節點F.................24
表 2.18  Bit Counting節點F............24
表 2.19  FP-List節點E.................25
表 2.20  Bit Counting節點E............25
表 2.21  FP-List輸出結果..............26
表 2.22  建構FP-table.................32
表 2.23  選擇出各節點項目並計數.......33
表 2.24  計數各個商品項目.............33
表 2.25 列出集合之項目節點D 1.........34
表 2.26 列出集合之項目節點D 2.........35
表 3.1  原始資料庫欄位................40
表 3.2  可參考的交易商品項目欄位......40
表 3.3  TempDB_PROD欄位...............41
表 3.4  TempDB_PROD_Cnt欄位...........42
表 3.5  TempDB_PROD_Item欄位..........42
表 3.6  TempDB_UserSeq欄位............43
表 3.7  Apriori_Table欄位.............44
表 3.8  FB-Tree-Table欄位.............44
表 3.9  列出所有組合1.................57
表 3.10  列出所有組合2................57
表 3.11  列出所有組合3................58
表 4.1  交易資料庫分佈................77

[1]. Richard J. Roger, Michael W. Geatz, Data Mining: A tutorial-based primer, First edition, Boston, Addison Wesley, Chinese Version, 2003, pp.79-85

[2]. R. Argawal and R. Srikant, "Fast Algorithms for Mining Association Rules", Proc. of the 20th International Conference on Very Large Databases, Sep.1994, pp.478-499.

[3]. J. Han, J. Pei, Y. Yin, and R. Mao, “Mining Frequent Patterns without Candidate Generation: A Frequent-Pattern Tree Approach”, Journal of Data Mining and Knowledge Discovery, 2004, pp.53-87

[4].Fan-Chen Tseng, Ching-Chi Hsu, and Henry Chen, “Mining Frequent Closed Itemsets with the Frequent Pattern List.”, Proc. The 2001 IEEE International Conference on Data Mining, San Jose, California, USA, Nov.29–Dec.2, 2001.

[5]. Fan-Chen Tseng and Ching-Chi Hsu, “Creating Frequent Patterns with the Frequent Pattern List”, Proc. of Asia Pacific Conference of Data Mining and Knowledge Discovery, Hong Kong, 2001, pp.376-386

[6]. Xuequn S., K.U. Sattler, and I. Geist, “SQL Based Frequent Pattern Mining withFP-growth”, Lecture Notes in Computer Science, Springer Berlin/Heidelberg, Volume 3392/2005, Apr. 2005, pp.32-46

[7]. Xuequn, S., K. U. Sattler, and I. Geist, “SQL Based Frequent Pattern Mining without Candidate Generation”, Proc. of the 2004 ACM Symposium on Applied Computing, ACM Press New York, USA, 2004, pp.618-619  

[8]. R. Agrawal, Tomasz Imielinski and Arun N. Swami, "Mining Association Rules between Sets of Items in Large Databases", Proc. of the ACM SIGMOD Conference on Management of Data, 1993, pp.207-216

[9]. Pramudiono I., T. Shintani, T. Tamura and M. Kitsuregawa, “ Parallel SQL based associaton rule mining on large scale PC cluster: performance comparision with directly coded C implementation”, Proc. of Third Pacific-Asia Conf. on Knowledge Discovery and Data Mining, 1999, pp.94-98