High-utility itemset mining is the task of discovering high-utility itemsets, i.e. sets of items that yield a high profit in a customer transaction database. High-utility itemsets are useful, as they provide information about profitable sets of items bought by customers to retail store managers, which can then use this information to take strategic marketing decisions. An inherent limitation of traditional high-utility itemset mining algorithms is that they are inappropriate to discover recurring customer purchase behavior, although such behavior is common in real-life situations (for example, a customer may buy some products every day, week or month). In this paper, we address this limitation by proposing the task of periodic high-utility itemset mining. The goal is to discover groups of items that are periodically bought by customers and generate a high profit. An efficient algorithm named PHM (Periodic High-utility itemset Miner) is proposed to efficiently enumerate all periodic high-utility itemsets. Experimental results show that the PHM algorithm is efficient, and can filter a huge number of non periodic patterns to reveal only the desired periodic high-utility itemsets.
展开▼