MINING HIGH AVERAGE UTILITY ITEMSETS WITH CORRELATION CONSTRAINT

Abstract

High utility itemset is a crucial concept in the itemset mining problem. To address the imbalance in utility between elements within an itemset, the concept of average utility has been proposed. However, some itemsets with high average utility still exhibit low element correlation, reducing their value for business analysis. To overcome this limitation, this paper proposes a method for mining high average utility itemsets that considers correlation through the CHAU (Correlated High Average Utility) algorithm. The research focuses on improving the formula for calculating the upper bound of average utility to increase candidate pruning capability, thereby improving the algorithm's processing performance. Experimental results comparing the proposed method with the state-of-the-art CoHAI algorithm across datasets with varying sparsity and density, including Chainstore, Kosarak, Retail, Accident, Mushroom, and Chess, show that the proposed method achieves better performance in both execution time and memory consumption.