An Efficient Algorithm for Mining Frequent Itemsets and Association Rules

Abstract

This chapter considers solving the problem of Frequent Itemsets Mining (FIM) in large-scale databases, which is known to be a subset problem with a complexity of order 2ⁿ. Despite extensive research related to this problem, however, proposed algorithms still suffer the problem of low performance in terms of execution times and main memory usage. In this chapter, we propose a binary-based approach towards solving the FIM problem. The proposed approach utilizes a binary representation of the database transactions as well as binary operations to ease the process of identifying the frequent patterns as well as reduce the memory consumption. Extensive computational experiments have been conducted using different publicly available datasets with different characteristics to test and benchmarking the performance of the proposed algorithm. The obtained results showed that the proposed binary-based approach outperforms current algorithms, achieving less execution time while also maintaining low memory usage.