| Data mining has been being a very active field of research. Its researchachievements are widely used in economics, management and other fields so as topromot the development of human society effectively and bring great benefits to theeconomy and society. Association rule data mining is the most important branch of datamining algorithm.However, the research and application of association rule data miningtechnology is far from extensive and profound at present, especially in the research ofassociation rule data mining with the huge amounts of data.This paper proposes an effective method on data mining of association rule withmassive data through researches on data mining methods with massive data. Thismethod has realized an optimal partitioning of the massive data and an overall miningof the massive data with association rule.As for the method of data partition, this paper utilizes particle swarm optimizationalgorithm for spatial clustering to optimize the division of massive data so that themassive data is divided into many small sub-databases. This study gives an introductionand makes an analysis of particle swarm optimization algorithm and clusteringalgorithm, summarizes the relevant improvement of the particle swarm optimizationalgorithm, and focuses on the study of how the data record code is converted intoparticles, how to establish fitness function, and how to avoid particle swarmoptimization algorithms falling into local convergence and so on. By adopting a fitnessfunction which meets the requirements of short-distance within-class and long-distancebetween-class, this research has greatly improved the accuracy of classification. At theend, details of spatial clustering algorithm based on particle swarm optimizationalgorithm.As regards the data mining of association rules, through the study of the classicassociation rule mining algorithm—Apriori algorithm, this paper has found that thealgorithm has the shortcomings of generating a large number of candidate sets andrepeated scan of the database, and is in a very low efficiency when dealing with massivedata, or simply unrealistic. Thus, this paper adopts an improved Apriori algorithm basedon a matrix of bit storage, which only scans database once, and is highly efficient byadopting “and†operations in the process of getting frequent item sets.With the actual experimental data and methods of data mining proposed by thecurrent research, the experiment results show that the proposed association rule datamining method based on particle swarm optimization algorithm is effective, which has not only inherited the advantages of the improved Apriori algorithm based on a matrixof bit storage, but also has solved the problem that Apriori algorithm is not suitable forassociation rule data mining with massive data. |
