<![CDATA[This research aims to compare the performance of the Apriori and FP-Growth algorithms in the process of data mining association patterns in the online sales transaction data of a bookstore. The dataset used consists of 74.090 transactions resulting from data cleaning from the period January-June 2025. The analysis was conducted through the stages of data collection, followed by data preparation consisting of data cleaning and data transformation, and then continued to the modeling stage of the two algorithms. The results of the experiment show that Apriori tends to be faster on small-scale datasets with simple transaction patterns, while FP-Growth has more stable memory usage and shows more efficient processing time on parameters that analyze larger data. Both algorithms produce identical numbers and contents of association rules for each parameter variation, indicating that the significant difference lies in performance efficiency, and not in the knowledge patterns produced. Rules with the highest lift values, such as the association between the books "Rumah Kaca" and "Jejak Langkah" (lift: 183,306 & confidence 0,903) and between the books "Namaku Alam" and "Pulang" (lift: 34,062 & confidence: 0,51) indicate strong purchasing patterns between titles with the same author and theme. These findings have the potential to support cross-selling strategies and product recommendations in online sales systems. This research is still limited to a relatively small and homogeneous dataset, so further using a broader data coverage is recommended to test the algorithm's performance more comprehensively.]]>
