DHUI: A new algorithm for mining high utility itemsets

• Published in: 2009 International Conference on Machine Learning and Cybernetics Vol. 1; pp. 173 - 177
• Main Authors: Chuan-Xiang Ma, Jie Zhang
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.07.2009
• Subjects: Association rules; Costs; Cybernetics; Data mining; High utility mining; Itemsets; Machine learning; Machine learning algorithms; Mathematics; Pattern growth; Testing; Transaction databases
• ISBN: 9781424437023; 1424437024
• ISSN: 2160-133X
• DOI: 10.1109/ICMLC.2009.5212570

High utility itemsets mining is another hot topic besides association rule in data mining field. It considers not only item's support values but also the quantities and profit. In current research, all algorithms mining high utility itemsets are based on candidate set generation-and-test category where too many candidates will be generated and tested. In this paper, an algorithm-DHUI based on PHUI-Tree structure is proposed. Using pattern growth approach mines high utility itemsets in PHUI-Tree without candidate generation, and calculates the utility of itemsets without scanning database repeatedly. The results of experiment show that DHUI is efficiency.