Application of Apriori and FP-Growth Algorithms in Analyzing Drug Purchasing Patterns

Abstract

Pharmacy sales transaction data contain valuable information on customer purchasing patterns; however, in practice, such data are often used merely as operational records, making relationships between purchased drugs difficult to identify. This study analyzes drug purchasing patterns using the Apriori and FP-Growth algorithms based on sales transaction data from Apotek Gadi Lamba Condet for the period January to June 2025. The transaction data were processed through data cleaning, drug name standardization, and transformation into transaction format, resulting in 7,038 transactions with 1,495 drug items. Association rule mining was performed using a minimum support of 0.01 and a minimum confidence of 0.17. The results show that the Apriori and FP-Growth algorithms generate ten identical association rules with the same support, confidence, and lift values, and all rules have lift values greater than one. Paracetamol 500 MG emerges as the most frequently involved drug in the association rules. These findings demonstrate that, for medium-scale pharmacy transaction datasets, Apriori and FP-Growth have equivalent capability in identifying drug purchasing patterns, with the primary difference lying in computational efficiency rather than the quality of the generated patterns.