FP-Growth Algorithm for Association Model Optimization in Household Sales Data

This research aims to find the value of support and confidence parameters needed so that associations between products can be identified and get the value of support, confidence, lift for the association rules found, and identify products that have the highest support value in frequent itemsets. The method used is Knowledge Discovery in Databases (KDD) with the stages of data collection, data pre-processing, data transformation, data mining, dan interpretation and evaluation. Sales transaction data was collected from January 1 to September 30, 2024, focusing on support and confidence values. The results showed that the association was successfully found with a parameter value of support 0.02 and confidence 0.5. In the association found, the products SWEAT BRONZE PANTS MINI M5 and SWEAT BRONZE PANTS MINI L5 have a support value of 0.004, confidence of 0.073, and lift of 1.421. These values indicate that although the frequency of this association is low, its strength exceeds that of a random association, which can be used in marketing strategies like product bundling.The product “SENSI PEREKAT S20” has the highest support of 0.149 (14.9%. The findings provide insight into the use of data mining algorithms to design data-driven marketing strategies and more efficient inventory management.