High On-Shelf Utility Mining Using an Improved HOUI-Mine Algorithm

Abstract

One of the recent upcoming data mining tasks, is to retrieve high profitable itemsets from a transaction database and is termed as High Utility Itemset (HUI) mining. Recently, numerous effectual algorithms were proposed in utility mining. High Utility Itemset mining considers the individual earnings of each items and the amount of items sold. On-shelf Utility Itemsets mining is one of the forms of HUI that considers the time period of items kept on the shelf of retail stores. This is the core gain feature of On-Shelf High Utility Itemsets. There are many On-Shelf utility mining algorithms are available. In this work, an existing HOUI-Mine algorithm was improved to retrieve On-Shelf HUI more quickly and efficiently. This improved algorithm was compared with existing HOUI-Mine algorithm. The algorithms were compared using two benchmark datasets, they are Mushroom and Chess. The execution time, memory space occupied and the number of itemsets retrieved are considered as performance factors. Experimental result shows that the improved HOUI-Mine algorithm performs better than HOUI-Mine.