An FP-split method for fast association rules mining

Recently, most of the studies on association rules mining focused on improving the efficiency of frequent itemsets generation. To our best knowledge, the FP-growth algorithm, which is based on the FP-tree to generate frequent itemsets is time-efficient. Currently, relevant studies are introduced to improve the FP-growth algorithm. However, they ignore the fact that the FP-tree construction may spend much time. Therefore, the goal of our research is to propose a fast algorithm called frequent pattern split, simply FP-split, for improving the process of the FP-tree construction. The proposed FP-split algorithm contains two main steps. The first step is to scan a transaction database only once for generating equivalence classes of frequent items. The second step is to sort these equivalence classes of frequent items in descending order so as to construct the FP-split tree. Through detailed experimental evaluations under various system conditions, our method shows excellent performance in terms of execution efficiency and scalability.