在現今競爭激烈與變化快速的環境下，瞭解與滿足顧客的需求是企業獲利的關鍵因素。因此，顧客關係管理就成為當今企業非常重視的議題。利用資料探勘技術可以幫助企業從龐大且複雜的顧客資料庫中找出隱含的資訊，其中又以關聯法則挖掘法最為著名。如何有效地推導出關聯法則，在過去已經有許多方法相繼被提出，但大部分的演算法對於(a)數值型資料的處理、(b)漸進式挖掘以及(c)線上挖掘等問題無法有效地同時加以處理。因此本研究提出以模糊切割與資料方塊為基礎之關聯法則演算法來建構單層次以及多階層等關聯法則演算法。並應用於顧客關係管理上的探討，來幫助企業在制定客製化行銷策略的決策上的一個有力的參考依據。 In the era of great competition, understanding and satisfying customers’ requirements are the critical tasks for a company to make a profit such that customer relationship management becomes the important business issue at present. With the help of the data mining techniques, the manager can explore and analyze from a great quantity of data to discover meaningful patterns and rules. Mining association rules from transaction databases is most commonly seen in data mining. However, most conventional algorithms can not simultaneously and effectively satisfy the following requirements: (a) the relationships among transactions with numeric values, (b) incremental mining, and (c) on-line mining. In this thesis, we integrate the data cube and fuzzy partition techniques to propose a single-level association rule miner and a multi-level association rule miner. This mined knowledge can be applied in customer relationship management to help decision marker make correct business decisions for marketing strategies.
