Comparison of Apriori and FP-Growth Algorithms in Analyzing Association Rules

Abstract

The problem objectives of this research include the following: To implement Apriori and FP-Growth Algorithms in determining the comparison of association rules and To build a jupyter notebook application model in determining the comparison of association rules of Apriori and FP-Growth Algorithms. This research compares Apriori and FP-Growth algorithms in analyzing association rules, with a focus on implementation and model development in Jupyter Notebook. Through manual calculation using 10 transaction data samples and testing on 38,765 groceries data entries from Kaggle, differences were found in the lift results between itemsets. Apriori algorithm often shows a negative relationship between items, while FP-Growth gives a similar interpretation but with slightly different lift values, showing a different influence in the relationship between items. In addition, FP-Growth proved to be more efficient with a much faster execution time (5.2757 seconds) than Apriori (185.9585 seconds), especially in handling large datasets. The results of this study indicate that the selection of an appropriate algorithm should consider the characteristics of the dataset and the purpose of the analysis.