A Bottom-Up Projection Based Algorithm for Mining High Utility Itemsets

Erwin, A., Gopalan, R.P. and Achuthan, N.R.

    Mining High Utility Itemsets from a transaction database is to find itemsests that have utility above a user-specified threshold. This problem is an extension of Frequent Itemset Mining, which discovers itemsets that occur frequently (i.e. with occurrence count larger than a user given value). The problem of finding High Utility Itemsets is challenging, because the anti-monotone property so useful for pruning the search space in conventional Frequent Itemset Mining does not apply to it. In this paper we propose a new algorithm called CTU-PRO that mines high utility itemsets by bottom up traversal of a compressed utility pattern (CUP) tree. We have tested our algorithm on several sparse and dense data sets, comparing it with the recent algorithms for High Utility Itemset Mining and the results show that our algorithm works more efficiently.
Cite as: Erwin, A., Gopalan, R.P. and Achuthan, N.R. (2007). A Bottom-Up Projection Based Algorithm for Mining High Utility Itemsets. In Proc. 2nd International Workshop on Integrating Artificial Intelligence and Data Mining (AIDM 2007), Gold Coast, Queensland, Australia. CRPIT, 84. Ong, K.-L., Li, W. and Gao, J., Eds. ACS. 3-10.
pdf (from crpit.com) pdf (local if available) BibTeX EndNote GS