IMPLEMENTATION OF ASSOCIATION RULE MINING TO ANALYZE PRODUCT PURCHASING PATTERN IN AN IRON AND STEEL COMPANY

Abstract

Companies operating in the steel and iron industry generate large volumes of sales transaction data on a daily. However, these data are often utilized solely for archival purposes, leaving the valuable information contained within them underutilized in supporting business decision-making. The proposed method implements the Association Rule Mining technique using the FP-Growth algorithm in RapidMiner on a sales transaction dataset collected from January 1, 2026, to June 4, 2026. The dataset consists of 143 transactions across 33 product item categories. The objective of this study is to analyze customer purchasing patterns and identify associations among products that are frequently purchased together. The research adopts the Cross-Industry Standard Process for Data Mining (CRISP-DM) methodology, which consists of business understanding, data understanding, data preparation, modeling, evaluation, and deployment. The process of generating association rules was carried out using the FP-Growth algorithm based on predetermined minimum support and minimum confidence thresholds. The results of the study generated several association rules that reveal relationships among steel and iron products, using a minimum support threshold of 10%, confidence = 100%, and a lift ratio ≥ 1, all of which met the specified criteria. The findings indicate that the four strongest association rules are: the purchase of hollow plafond is always followed by the purchase of spandex screws; the combination of Alderon Spandex and hollow plafond is always followed by the purchase of spandex screws; the combination of Dynabolt and hollow plafond is always followed by the purchase of spandex screws; and the combination of Alderon Spandek, Dynabolt, and hollow plafond is always followed by the purchase of spandex screws. Each of these rules achieved a confidence value of 100% and a lift value of 1.192. Meanwhile, light steel roofing screws were identified as the most frequently occurring product in the transaction data, with a support value of 0.958 (95.8%). The knowledge extracted from these patterns can be utilized to support the development of sales strategies, optimize product placement, improve inventory management, and design more effective promotional packages. Therefore, the implementation of Association Rule Mining successfully transforms transactional data into valuable knowledge, enabling more accurate decision-making and enhancing the company's operational effectiveness