快速模組拆解之關聯規則探勘-QMD

近年來，客戶關系管理(CRM)是個相當熱門的議題，因爲企業必須瞭解消費者購物行爲與商品間的關聯關系，才能妥善安排商品陳列順序。如此可以提升客戶滿意度，减少購物的搜尋時間。再者可以刺激購買商品數量，用以增加企業的利潤。所以在大型交易資料庫中，利用資料探勘技術找出的關聯法則，來提供企業的决策支援是非常重要的。本研究提出新的演算法QMD (Quick Modulized Decomposition)來找出商品間的關聯法則。QMD演算法的優點如下：1.只需掃描資料庫一次；2.利用模組化方式來提升執行效率；3.利用遮罩(mask)與布林模式(Boolean)來産生拆解項目因子模組。上述得知，透過本演算法做關聯分析，其效能將優於以往Apriori-Base的演算法。此外，關聯法則的推導過程中，將不會重複産生多餘的候選項目組，因此更勝於拆解模式的演算法。快速得到正確、有效用的資訊，是企業在數位時代中最大的利器，由此能降低時間成本、快速反映市場需求，是提升競爭力的最大利基。

關鍵字

資料探勘；關聯法則； Apriori演算法；高頻項目集

並列摘要

Recently Customer Relationship Management is on of the hottest issues in cooperations. In order to proerly arrange the position of products, Cooperations need to understand customers' shopping behaviors and the associationships between products. In this way, we can increase the customers' satisfication and decrease the searching time during shopping. Besides, we can increase the quantity of purchase products and the profits. Thus, it is very important to use the technology of data mining to find the useful association rules and to provide the cooperation's decision supports. In this paper we propose a new algorithm QMD (Quick Modulized Decomposition) to find the association rules from large transaction databases. The merits of QMD algorithm are: 1. In data mining process it only needs to scan whole transaction database once. 2. Using Modulized method to increase the performance of data mining process. 3. Using mask and Boolean method to decompose the itemsets to sub-itemsets. 4. In decompostion process, we combine the same sub-itemsets and get the supports of each sub-itemset very efficiently and significantly shorten the processing time and cost.

並列關鍵字

Date Mining ； Association Rule ； Apriori ； Frequent itemsets

參考文獻

Agrawal R.,Imilienski T.,Swami A.(1990).Mining Association Rules between Sets of Items in Large Databases.(In Proc. of the ACM SIGMOD Int`l Conf. on Management of Data).

Google Scholar

Agrawal R.,Srikant R.(1994).Proc. 1994 In`t Conf. VLDB.Santiago, Chile:

Google Scholar

Brin S.,Motwani R.,Silverstein C.(1997).Beyond Market Baskets: Generalizing Association Rules to Correlations.1997 ACM SIGMOD Conference on Management of Data.(1997 ACM SIGMOD Conference on Management of Data).

Google Scholar

Brin S.,Motwani R.,Ullman J.D.,Tsur S.(1997).Dynamic Itemset Counting and Implication Rules for Market Basket Data.ACM SIGMOD Conference on Management of Data.(ACM SIGMOD Conference on Management of Data).