快速反向關聯法則與調整緊密規則－促銷商品組合之應用

Fast-Backward Association Rule and Update-Compact Rule-Determination of Purchasing Patterns

蔡玉娟(Yuh-Jiuan Tsay) ； 張簡雅文(Ya-Wen Chang Chien) ； 黃彥文(Yen-Wen Hung)

《資訊管理學報》 10卷1期 (2003/07) Pp. 181-204

https://doi.org/10.6382/JIM.200307.0181

摘要

企業藉由所建立之專屬會員制度，透過資料探勘技術從龐大的會員交易資料庫發掘消費特徵，實現個人化之服務，有效區隔市場與訂定行銷策略。資料探勘技術之關聯法則的執行程序受限於必須由單一項目集，逐層擴展，經過長時間之重複組合與運算步驟，才能發掘合適之高頻項目集。本研究提出一個新的快速反向關聯法則(Fast-Backward Association Rule, FBAR)以克服上述之缺點，並應用於發掘特定促銷項目之商品組合。FBAR之執行程序反向於Apriori關聯法則，執行步驟為：(1)建立促銷目標資料表(table)一婦描交易資料庫一次，將交易資料庫中不符合預定促銷的項目刪除，而保留符合預定促銷的項目，記錄在促銷目標資料表並暫存於主記憶體；(2)分解促銷目標資料表之交易資料－在促銷目標資料表中，由最長交易資料開始逐層分解項目集；(3)發掘符合最小支持度之高頻項目集－當分解至某長度之項目集且已符合最小支持度，則停止該項目集之分解。FBAR僅需掃描資料庫一次，而將刪減之交易資料記錄在促銷目標資料表並暫存於主記憶體，再透過分解較長交易紀錄，可快速發掘符合最小支持度之高頻項目集。藉由調整緊密規則法(Update-Compact Rule, UCR)可轉換高頻項目集為涵蓋率與緊密程度較高之關聯規則。

關鍵字

資料探勘 ； 關聯法則 ； 快速反向關聯法則

並列摘要

The discovery of association rules is an important and popular data-mining task, for which many algorithms have been proposed. These solutions are flawed, containing weaknesses that include often requiring repeated passes over the database, and generating a large number of candidate itemsets. In order to overcome the bottlenecks of association rule, we propose a new Fast-Backward Association Rule (FBAR) which is used to discover the specific promotional bundles. The procedure of FBAR is against Apriori algorithm. Efficiency of mining is achieved with three steps: (1) establish promotional table with specific promotional items-delete non-promotional items from database; (2) decompose transactions of promotional table-decomposing from the longest transaction record in promotional table, level by level; (3) find out frequent patterns conform to the minimum support-algorithm terminates when the calculated support is greater than, or equal to, the minimum support. The Update-Compact Rule (UCR) algorithm is obtained by modifying Compact Rule Set to reason the knowledge rules from frequent patterns. FBAR not only prunes considerable amounts of data reducing the time needed to perform data scans and requiring less memory, but also ensures the correctness of the mined results.

並列關鍵字

Data Mining ； Association Rule ； Fast-Backward Association Rule

參考文獻

Agrawal, R.,Imielinski, T.,Swami, A.(1993).Proceedings of the 1993 ACM SIGMOD International Conference on Management of Data.Washington:

Agrawal, R.,Srikant, R.(1994).Proceedings of the 20th International Conference on Very Large Databases.Santiago:Chile.

Agrawal, R.,Srikant, R.(1995).Proceedings of the International Conference on Data Engineering (ICDE).Taipei, Taiwan:

Berry, M. J. A.,Linoff, G. S.(1997).Data Mining Techniques: For Marketing, Sales, and Customer Support.New York, NY:John Wiley ＆ Sons, Inc..

Berzal, F.,Cubero, J. C.,Marin, N.,Serrano, J. M.(2000).TBAR: An efficient method for association rule mining in relational databases.Data ＆ Knowledge Engineering.37(1)