Finding Frequent Items: A Novel Method for Improving the Apriori Algorithm

Abstract

In the current paper, we use an intelligent method for improved the Apriori algorithm in order to extract frequent itemsets. PAA (proposed Apriori algorithm) is twofold. First, it is not necessary to take only one data item at each step. In fact, all possible combinations of the items could be generated at each step. Secondly, we can scan only some transactions instead of scanning all the transactions to obtain frequent itemset. For performance evaluation, we conducted three experiments with the traditional Apriori, BitTableFI, TDM-MFI, and MDC_Apriori algorithms. The results exhibit that due to the significant reduction in the number of transaction scans to obtain the itemset, the algorithm execution time is significantly reduced; as in the first experiment, the time spent to generate frequent items underwent a reduction by 52% compared to the algorithm in the first experiment. In the second experiment, the amount of time spent is equal to 65%, while in the third experiment, it is equal to 46%.