<![CDATA[This study aims to optimize the association rule mining process on retail transaction data by applying the Binary Particle Swarm Optimization (BPSO) algorithm. Classical methods such as Apriori and FP-Growth often face efficiency limitations, particularly when dealing with large-scale datasets, due to repeated candidate generation processes and high memory requirements. BPSO is employed as a metaheuristic approach capable of adaptively exploring the search space through binary itemset representation and a fitness function based on support, confidence, and lift values. This research follows the CRISP-DM framework, encompassing the stages of data understanding, data preparation, modeling, and evaluation. Based on retail transaction data from KLM, the BPSO process produced the best particle containing 16 potential itemsets. The calculation of support and confidence resulted in seven item combinations that met the minimum threshold and generated eighteen association rules. Evaluation using the lift ratio showed that all rules have lift values greater than one, indicating strong and meaningful relationships among products. These findings demonstrate that BPSO is effective in discovering relevant association patterns and can support retail decision-making, such as product arrangement, cross-selling strategies, and the development of recommendation systems.]]>
Copyrights © 2026
