A Hybrid algorithm for mining high utility itemsets from transaction databases with discount notion

Abstract

High-utility itemset mining has attracted signicant attention from the research community. Identifying high-utility itemsets from a transaction database can help business owners to earn better profit by promoting the sales of high-utility itemsets. The technique also finds applications in web- click stream analysis, biomedical data analysis, mobile E-commerce etc. Several algorithms have been proposed to mine high-utility itemsets from a transaction database. However, these algorithms assume that items have a constant profit associated with them and don't embed the notion of discount into the utility-mining framework. In this thesis, we integrate the notion of discount in state-of-the-art utility-mining algorithms and propose a hybrid- algorithm for efficient mining of high-utility itemsets. We conduct extensive experiments on real and synthetic datasets and our results show that our proposed algorithm outperforms the state-of-the-art algorithms in terms of total execution time and number of itemsets that need to be explored.