Unveiling Purchase Patterns: A Comparative Analysis of Apriori, FP-Growth, and Eclat Algorithms in Market Basket Analysis

Market Basket Analysis (MBA) is a pivotal data mining technique aimed to discover the invaluable insights into consumer behavior through the exploration of transactional data in the retail market. This paper presents a comparative performance analysis of three prominent MBA algorithms: Apriori, FP-growth, and Eclat these are widely used for extracting frequent itemsets and generating association rules from retail transactional database. The study begins with a comprehensive overview of each algorithm with their fundamental principles and methodologies. It then proceeds to evaluate their performance across various metrics such as execution time, memory usage, and scalability using real-world transactional datasets. The dataset of this research has been generated from the transaction slips of a retail supermarket in Bangladesh. The experiments are conducted using different sizes of transactional data to assess how each algorithm handles varying data volumes. Special attention is given to the strategies employed by each algorithm to optimize the mining process and reduce the computational overhead. Our findings indicate that, the FP-growth based mining approach outperforms the Apriori and Eclat algorithms in terms of runtime efficiency and memory utilization, especially while the size of the dataset increases. For our transactional dataset of 100000 transactions alongside 2382 unique items, the Fp-growth algorithm beat the Apripri and Eclat in execution time by 4 and 1.87 times, respectively. However, the choice of algorithm can depend on specific dataset characteristics and the desired balance between mining speed and resource consumption.