Apriori-Based Association Rule Mining Approach for Developing a Product Recommendation System in an Agricultural E-Marketplace

The rapid growth of transactions in Indonesian agricultural marketplaces leaves a gap between large data volumes and suboptimal marketing strategies. As a novel contribution, this study explicitly applies the Apriori algorithm to the context of agricultural e-marketplaces in Indonesia, serving as a validated case study. The primary data used in this analysis is a collection of historical transaction data obtained from one such marketplace. The dataset described in this case study includes 10 transaction histories involving 22 product items. This study aims to transform untapped historical data into business-strategy insights. By setting minimum support and confidence to 50%, the analysis successfully identified significant association rules. The strongest rule indicated a co-purchase pattern between specific products with a confidence value of 60.6% and a Lift Ratio of 11.1, indicating a robust positive correlation. This rule was then successfully implemented into a functional recommendation feature. Validation testing demonstrated complete consistency between the system results and manual calculations. This case study demonstrates the effectiveness of Apriori and provides a benchmark for developing similar technologies to improve sales and user experience in Indonesia's digital agriculture sector.