Research And Application Of K-medoids Clustering Algorithm Based On ?_o-neighborhood Search Strategy

With the rapid development of computer technology and increasing expand needs of humans,massive data is piled up in this era.In order to extract useful knowledge and information from massive data,data mining comes into being.Cluster analysis as an important research branch of data mining,is aim to divide the data into meaningful or useful clusters by analyzing the similarity among the data.K-medoids algorithm is a one of the classical clustering algorithm,which is not sensitive in the database of containing isolated point or noise and has a good stability,so widely used.But traditional K-medoids clustering algorithm is sensitive to the initial centers of random selection and has a slow convergence speed and gets not high clustering accuracy.In view of the traditional K-medoids clustering algorithm above shortcomings,this paper proposes a feasible improved algorithm to overcome its shortcomings.In this paper,the main research work is as follows:In order to solve the problem that traditional K-medoids clustering algorithm is sensitive to the initial selection of the centers,this paper establishes a ?0-area block for each object of the database and selects K ?0-areas which their densities are larger and the distances are far away for each selected ?0-areas blocks,puts the core objects of the corresponding ?0-areas blocks as the K initial centers;updates K centers by using ?0-area block search strategy to reduce the number of iterations.What's more,this paper presents a weighted criterion function based on between-clusters distance and within-clusters distance to improve clustering accuracy.The results of experiments show that this improved algorithm tested in standard data set Iris and Wine of UCI,get ideal initial centers located in difference clusters,find the optimal solution in less iteration,and improve the accuracy of clustering algorithm greatly.Finally,making this improved K-medoids clustering algorithm apply into the intrusion system,we obtain the ideal results and further prove the effectiveness and superiority of the algorithm in this paper.