Association rule mining (ARM) is a technique within data analysis that focuses on identifying frequent patterns or itemsets from datasets. Several algorithms, such as Apriori and FP-Growth, have been developed to perform this task. In this paper, we explore a selection of well-established ARM algorithms, including Apriori, Frequent Pattern Growth (FP-Growth), Equivalence Class Clustering with a bottom-up Lattice Traversal (ECLAT), Rapid Association Rule Mining (RARM), and Enhanced Apriori. These algorithms represent both classic and modern approaches in ARM, offering a diverse spectrum of methodologies for frequent pattern extraction. This study aims to assess and compare the effectiveness of these algorithms on a real-world dataset of varying sizes. We utilize the online retail dataset from the UCI machine learning repository dividing it into four subsets for analysis. Each algorithm is tested on these subsets, and its performance is evaluated using several criteria, such as execution time, memory consumption, quality of generated rules, and the total number of rules produced. Furthermore, we assess the impact of different support thresholds on the quantity and quality of rules generated to provide a thorough comparison across all scenarios. The results show that Enhanced Apriori reduced execution time by 48% and memory consumption by 43% compared to Apriori. FP-Growth showed 43% faster execution time for small datasets and 63% faster performance for larger datasets, while consuming 45% less memory on small datasets and 50% less memory on larger datasets. Additionally, ECLAT performed 62% faster on small datasets and 47% faster on larger datasets, with 57% and 53% lower memory consumption, respectively. These findings highlight significant efficiency improvements in both runtime and memory usage, especially for larger datasets, and provide valuable insights into selecting algorithms based on dataset characteristics and ARM application requirements.