Developing a novel H-Apriori algorithm using support-leverage matrix for association rule mining

The importance of e-commerce platforms is playing a vital role in human life. Data mining plays a crucial role in e-commerce platforms, allowing them to extract valuable information from extensive datasets by examining buying tendencies from customers using Association rule mining and Market basket analysis. In this research paper, the author proposed a heuristic H-Apriori algorithm built on the support-leverage of an item and the effective weight of a transaction using a single dataset scan. The trial estimation contains five sample transactional datasets with variable sizes, ranging from 20 to 7500 transactions. The performance of H-Apriori algorithm is compared with state-of-the-art algorithms, Improved-A priori, Hybrid-Apriori, and classic Fp-Growth. Improved-Apriori and Hybrid-Apriori are relying on queue data structure, while classic Fp-Growth is based on a tree data structure. The results of this study demonstrate that the proposed heuristic H-Apriori algorithm outperforms the improved A priori, Hybrid-Apriori, and Fp-Growth with accuracy of frequent item set and interesting item set count, taking up to 99%,75%, and 93% less run time for frequent item set and up to 91% for interesting item set, as well as up to 91%,15%, and 13% less memory consumption.