Evaluasi Performa Algoritma FP-Growth Berdasarkan Variasi Parameter Minimum Support dan Confidence pada Dataset Groceries

Abstract

This research investigates the relationship patterns among products in the Groceries dataset by applying the FP-Growth algorithm as an approach to uncover association rules. The analysis was conducted by varying the values of minimum support and minimum confidence to observe how these parameters influence the number and quality of generated rules. The experimental findings reveal that the combination of a support value of 0.01 and a confidence value of 0.4 generated the largest number of rules, totaling 71, with the highest lift value reaching 2.344. These results indicate a strong association between several products that frequently appear together within a single transaction, where whole milk emerges as the most dominant item, both as an antecedent and as a consequent. A high lift value suggests that customers who purchase whole milk are more likely to buy related items such as yogurt, curd, or cream cheese. The insights from this study can serve as a valuable reference for retailers in designing more effective product placement, improving promotional strategies, and supporting data-driven business decisions, particularly in cross-selling and inventory optimization.