A Computer Simulation Study of an Improved Apriori Algorithm

Aiming at the inefficiency of Apriori algorithm, this paper proposes an improved Apriori algorithm, which is tested and evaluated in combination with computer simulation research. The improved algorithm aims to improve the efficiency and accuracy of association rule mining by introducing decision tree strategies and parallel processing methods. This study used multiple datasets for experiments and compared the performance indicators of the improved algorithm with the traditional Apriori algorithm. Firstly, an improved Apriori algorithm is proposed to reduce the number of candidates sets and reduce the time complexity of the algorithm through the decision tree strategy. The parallelization method further improves the efficiency of the algorithm by dividing the data set into subsets and using multiple threads for simultaneous processing. Then, multiple real and integrated datasets are used for experimental evaluation. The results show that the improved algorithm has significantly reduced the running time compared with the traditional Apriori algorithm under the same data set, and there is no obvious difference in the generated frequent itemset and association rules. This shows that the decision tree strategy and parallelization processing method can effectively improve the efficiency of the algorithm, which is suitable for the association rule mining task of largescale datasets.