A BPSO-based method for high-utility itemset mining without minimum utility threshold

Abstract

High-utility itemset mining is used to obtain high utility itemsets by taking into account both the quantity as well as the utility of each item, which have not been considered in frequent itemset mining. Many algorithms compute high utility itemsets by setting a minimum utility threshold in advance. However, determining the minimum utility threshold is not easy. Too high or too low a threshold may result in incorrect high utility itemsets. In this paper, we propose a method based on binary particle swarm optimization to optimize the search for high utility itemsets without setting the minimum utility threshold beforehand. Instead, the application of the minimum utility threshold is performed as a post-processing step. Experiments on five datasets indicate that the proposed method is better than existing methods in finding high utility itemsets, and the time to obtain those itemsets is faster than that with setting the minimum utility threshold first.