Comparison of Priori and FP-Growth Algorithms in Determining Association Rules

Abstract

Collection of transaction data on the sale of goods of a minimarket, which is increasing every day, is often treated only as a record, causing data to accumulate in a database. Data mining can help the decision-making process quickly, making it possible to manage the information contained in transaction data into new knowledge. The purpose of this study was to analyze the data of sales transactions of goods by comparing the Apriori algorithm and the FP-Growth algorithm, to determine the association rules based on consumer purchase patterns with association techniques that seek several frequent itemsets and proceed with the establishment of Association Rules. The research utilizes primary data in the form of sales transactions from February to March 2018. The results of analyzing goods sales transaction data using Apriori algorithm and FP-Growth algorithm by setting a minimum support value of 4% and a minimum value of confidence of 19% is to produce a number of rules different associations where the Apriori algorithm produces 11 rules while FP-Growth produces 10 rules but has the final association value (the same and the execution time required by the FP-Growth algorithm is faster with 0.5 seconds than the Apriori algorithm which takes 0.6 seconds. This information very useful for managing the layout of items close together and allows designing marketing concepts to also provide stock.