以利潤為主要考量之多重最小支持度量化關聯規則

Mining Quantitative Association Rules with Multiple Minimum Supports Based Mainly on Profits

鄭印良(Yin-Liang Cheng)

國立虎尾科技大學/管理學院/資訊管理研究所/碩士(2007年)

摘要

一般關聯規則的特性主要是考慮商品在交易中的關聯性，使得與重要商品相關的關聯規則能被挖掘出來；而挖掘量化關聯規則的主要目的，則是從交易資料庫中找出大部分的客戶購買了哪些數量的商品，也會同時購買哪些數量的其他商品。但是，一般關聯規則沒有考慮商品被購買的數量，而量化關聯規則也沒有考慮到商品本身的利潤，因此如何同時考慮商品本身的利潤與其被購買的數量，成為本研究的議題。本論文主要應用分割演算法(Partition algorithm)將其資料量化，之後建立成P-tree的結構，再加上ABC理論及多支持度的概念，提出了一個新PMSFP-tree（Profit and Multiple minimum Supports Frequent Patterns tree）結構與PMSQFP-growth(Profit and Multiple minimum Supports QFP-growth)演算法。

並列摘要

Association rules takes an important role in Data Mining. By applying this technique, the relations between the data in a transaction database can be found, and then managers utilize the explored information to make decisions.
The traditional association rules technique mostly focuses on the amount of trades, so that the goods with high profits and low sales volume will be neglected. This paper proposes a method using the concept of P-tree(Patterns tree), CFP-growth and QFP-growth. Subsequently, the common stock-checking method, ABC, is utilized to classify the goods so as to those goods with high profits and high support will be remained. Finally, the explored information will be much expected to conform to managers.
By combining the concepts as mentioned above, a novel technique, PMSFP-tree (Profit and Multiple minimum Supports Frequent Patterns tree) and PMSQFP-growth (Profit and Multiple minimum Supports QFP-growth) is proposed in this paper. Experimental results demonstrate that our method can improve the drawbacks of CFP-growth. Besides, by taking account of trade profits, the mined results will be much respond to its actual facts.

參考文獻

[28]Pauray S. M. Tsai, Chien-Mining Chen, “Mining quantitative association rules in a large database of sales transactions” , Journal of Inforamtion Science and Engineering, Vol.17 No.4, pp.667-681, 2001.

[2]R. Agrawal, T. Imielinksi, A. Swami, “Mining association rules between sets of items in large database”, In Proc. ACM-SIGMOD Conference Management of Data (SIGMOD’93), pp. 207–216,1993.

[3]Han J., Pei, Pei J.,Yin Y.,Mao R., Mining Frequent Patterns without Candidate Generation: A Frequent-Pattern Tree Approach, Data Mining and Knowledge Discovery, vol.1,no. 8, pp. 53-87, 2004.

[7]J. Roberto and Jr. Bayardo., “Efficiently Mining Long Patterns from Databases”, In Proc. of the ACM-SIGMOD Int'l Conf. on Management of Data, pp.85-93, 1998.

[8]Nicolas Pasquier, et al., “Efficient mining of association rules using closed itemset lattices”, Information Systems, Vol. 24, no.1, pp.25-46, 1999