Correlated High Average-Utility Itemset Mining

High average-utility itemset (HAUI) mining is an advancement over high utility itemset mining, where average-utility is used instead of utility measure to discover meaningful patterns. It has been discussed in several past studies that significance of utility-based patterns can be amplified if items in the patterns are correlated. In this paper, we propose a HAUI mining algorithm named correlated high average-utility itemset (CoHAI) miner which provides highly profitable and non-redundant correlated patterns. CoHAI miner is a one-phase algorithm which uses a vertical list structure to store utility and correlation information to apply pruning and candidate generation. Moreover, use of vertical list avoids repetitive dataset scans to calculate the utility of candidate itemsets. We performed rigorous experiments to compare the CoHAI miner with existing algorithm. The results show that the CoHAI miner performs efficiently and produces more meaningful patterns than existing algorithm.