Frequent Itemset Mining, or called FIM, is the most important task in association mining proposed by [1]. Given a set of transactions, a transaction is defined as a set of items have occurred together. For example, in a supermarker, when a custom buys milk, beef and beer. Her transaction is generated by the custom containing three item (milk, beef and beer). The goal of FIM tries to generate all the frequent-item whose freqeuncy are more than the predefined threshold. see Wikipedia for more detail information: http://en.wikipedia.org/wiki/Association_rule_learning
In this project, I have implemented apriori algorithm[1] and its variance called bitmap compression[2]. Both of these two version are built on Visual Studio project. Feel free to use and report errors for me, thanks.
If you need more efficient approaches, such as hmine, fp*, op or GPU-based approach, please mail to me, thanks.
how to use: fim.exe input-data-set minimul-support-value
and you can download more trial data for demonstrating from here http://www.iis.sinica.edu.tw/~hhyeh/fim/frequent\_itemset.html.
[1] Agrawal R, Imielinski T, Swami AN. "Mining Association Rules between Sets of Items in Large Databases." SIGMOD. June 1993, 22(2):207-16 [2] Liu Yang, Mei Qiao, "A Bitmap Compression Algorithm for Vertical Association Rules Mining," iscsct, vol. 2, pp.101-104, 2008 International Symposium on Computer Science and Computational Technology, 2008