A Dynamic Itemset Counting Based Two-Phase Algorithm for Mining High Utility Itemsets

Bhat, Anup B and Harish, S V and Geetha, M (2018) A Dynamic Itemset Counting Based Two-Phase Algorithm for Mining High Utility Itemsets. In: IEEE India Council International Conference (INDICON 2018), 16/12/2018, Amrita Vishwa Vidyapeetham.

[img] PDF
1280.pdf - Published Version
Restricted to Registered users only

Download (391kB) | Request a copy


High Utility Itemset Mining (HUIM) aids in the discovery of itemsets based on quantity and unit price of the items from a transactional database. Since its inception, HUIM has evolved as a generalized form of Frequent Itemset Mining (FIM). Unlike the support of an itemset which is antimonotone and is exploited in the algorithms for mining frequent itemsets, the utility measure is neither antimonotone nor monotone. This makes the problem of mining High Utility Itemsets (HUIs) interesting. In the current study, a novel method based on Dynamic Itemset Counting (DIC) has been proposed to optimize the Apriori-like Two-Phase (TP) algorithm for mining HUIs. Although, the TP algorithm uses antimonotonicity of Transaction WeightedUtility(TWU)ofitemsetstoprunethesearchspace,the candidates are generated in a level-wise manner. This requires multiple database scans to test the candidates. The proposed method tests and generates higher order candidates at different stops during the database scan and segregates the itemsets for further evaluation. Experiments performed on real-time datasets show significant improvement in execution time of the DIC method when compared to the TP algorithm

Item Type: Conference or Workshop Item (Paper)
Uncontrolled Keywords: High Utility Itemset Mining, Transaction Weighted Utility, Candidate Generation
Subjects: Engineering > MIT Manipal > Computer Science and Engineering
Depositing User: MIT Library
Date Deposited: 03 Jul 2019 03:59
Last Modified: 03 Jul 2019 03:59
URI: http://eprints.manipal.edu/id/eprint/154078

Actions (login required)

View Item View Item