Efficient Algorithms for Mining High Utility Quantitative Itemsets

  • 李 佳樺

Student thesis: Master's Thesis


High utility quantitative itemsets (abbreviated as HUQIs) mining is a novel research topic in data mining field which considers the concept of utility like quantity and profit to find high utility itemsets carrying information about quantity In market analysis it could reveal to decision-makers that shopping behavior could bring high profit to the company For example the customers who purchase M to N units of a product A are likely to purchase P to Q units of a product B However existing algorithms involve high computational cost and traditional HUQI mining may produce too many itemsets during the mining process causing performance problems and making results hard to be utilized by users In view of this this thesis proposes a novel algorithm named HUQI-Miner (High Utility Quantitative Itemsets Miner) for efficiently mining HUQIs in databases Moreover to efficiently present users concise mining results this thesis proposes two new types of HUQIs named concise HUQIs and summarized HUQIs respectively Concise HUQIs can be used to infer the HUQIs satisfying certain constrains while summarized HUQIs provide users a summarization about certain HUQIs Experimental results on both real and synthetic datasets show that the numbers of concise and summarized HUQIs can be several orders of magnitudes smaller than that of HUQIs Moreover HUQI-Miner outperforms the state-of-the-art algorithms in terms of both execution time and memory usage
Date of Award2015 Aug 14
Original languageEnglish
SupervisorSun-Yuan Hsieh (Supervisor)

Cite this