Association Rule Mining Research And Application Of The Algorithm

Association rule mining refers to the process of discovering the association relations between one item and another of a data set. Association rules , as one of the major branches of data mining technology, is gaining its attention at an ever-growing rate. In spite of this, the association algorithm of association rules can hardly meet the current computing requirements due to the constant change in data type and dimensions to be processed. It is, therefore, of great importance in terms of both theory and practice to find a more efficient algorithm of association rules.This paper, after probing into the classic algorithm of association rules, Apriori, and analyzing in detail the current types of algorithm, has presented the cluster subentry association rule(CSAR)algorithm that combines the clustering analysis method and association rules, based on the basic principle of association rule acquisition. This algorithm first focuses on the combines of the clustering method and association rule, use the subentry cluster sets as the leaves of the cluster tree, from the leaves we can have the optional rules. Second the subject of particle swarm optimization is put forward to work within the clustering algorithm and k-means algorithm combined together, which is efficient and convenient in multidimensional clustering method. Beside this we use the dynamic minimum support and seeds method to solve the effect which produced from the cluster. At last, by using the multi-dimensional constraint processing method to constrain or process the alternative rules, we can reduce the redundancy of output rules and the adverse effects to association rules.Through the experiment with different types of data verification, we can prove the feasibility and the advantage of the CSAR algorithm. From the experiment results compared with other algorithm, besides the fully excavate association rules, the advantage of this algorithm also shows in dealing with multidimensional data, improving mining speed and accuracy. In addition, because of the multidimensional constraint, the superiority of this algorithm also improves situations in which the rule discovering of large volume of data can not be easily analyzed and redundancy possess higher adverse impact. At last it improves the quality of the rules and in the actual digging, it increases core competitive ability in the process of decision making.