Study Of Data Mining Based On Multi-Marker Activation Algorithm And Genetic Algorithm

Data Mining is a new research field rising in recent years, and it is a hot research issuefor the current database system and its applications. Association Rules Mining and Clusteringis an important model for data mining, the relationship between the characteristics and thefactors of the individual can be found by association rules mining based on the results ofclustering. Genetic Algorithm is applied to the association rules mining because of itsexcellent robustness and global search capability. But the “premature” phenomenon and thedecline of convergence rate are impact on the efficiency of association rules mining. So it issignificant to improve the efficiency of association rules mining by improve the geneticalgorithm and its effective integration with the clustering algorithm. It is possible to combinethe genetic algorithm and the clustering algorithm by using matrix coding because alloperations of the genetic algorithm is based on real variables encoding, therefore an efficientclustering algorithm based on matrix is necessary.In this paper, we analyze and study the basic theories of data mining，association rulesand genetic algorithm, and analyzed the multi-marker propagation clustering algorithm. Amore efficient clustering algorithm based on weighted matrix—multi-marker activationclustering algorithm is proposed based on the thinking of multi-marker in this paper.Meanwhile, in this paper, the traditional genetic algorithm is improved based on the study ofgenetic algorithm, first of all, the intelligent determination method of support threshold isproposed, which we based on to improved the fitness function; Secondly, we proposed anddesigned an antibody operator base on the combination of biological immune mechanism,introduced the concept of the individual immunity; Finally, by using the antibody operator,the crossover operator and mutation operator in genetic manipulation of the traditional geneticalgorithm are improved. The practical examination shows that the improved genetic algorithmis effective and feasible in association rules mining.It can be show from the simulation that the performance of multi-marker activationalgorithm proposed in this paper is more superior, and the improved genetic algorithm in thispaper is significantly improved in terms of accuracy of the optimal solution, solving accuracyand astringency.In the future study, the method of making a weighted matrix in multi-marker activationclustering algorithm will be researched and studied, and how to improve the immune systemof individual and antibody operator will be researched too.