Investigating Crossover Operators in Genetic Algorithms for High-Utility Itemset Mining

Abstract

Genetic Algorithms (GAs) are an excellent approach for mining high-utility itemsets (HUIs) as they can discover most of the HUIs in a fraction of the time spent by exact algorithms. A key feature of GAs is crossover operators, which allow individuals in a population to communicate and exchange information with each other. However, the usefulness of crossover operator in the overall progress of GAs for highutility itemset mining (HUIM) has not been investigated. In this paper, the headless chicken test is used to analyze four GAs for HUIM. In that test, crossover operators in the original GAs for HUIM are first replaced with randomized crossover operators. Then, the performance of original GAs with normal crossover are compared with GAs with random crossover. This allows evaluating the overall usefulness of crossover operators in the progress that GAs make during the search and evolution process. Through this test, we found that one GA for HUIM performed poorly, which indicates the absence of well-defined building blocks and that crossover in that GA was indeed working as a macromutation.