Hierarchical Clustering Algorithm And Applications

Clustering analysis is a major field in data mining ,which is an important method of data partition. Clustering analysis now has been applied into various ways in statistics,machine learning,spatial database,biology and market analysis and so on. ClusteringAlgorithms includes partitioning,hierarehieal,density-based, grid-based, model-based Algorithm.The hierarchical clustering Algorithm and the partitioning Kmeans Algorithm are used frequency because of its simple use and its handle large data ability.Focus on the hierarchical clustering Algorithm's high compute,this paper through sorting the distance of the data to avoid recalcuting the distance between the class, thus lowered the calculate complex,it make the flexible of the Algorithm better. In order to avoid the hierarchical clustering Algorithm's cannot adjust the data's class,we take over the class strucster,attain the class Centroid, which can be used as the Kmeans Algorithm initial class centroid. Thus avoided the result unsteady caused by the Kmeans Algorithm initial class Centroid select from random.In the last of this paper, we propose a method of design RBF Neural Network using the improved Clustering algorithm to determine the number and center and spread of Redial Basis Function,its output layer's weights determines by the linear equations. The simulate experiment of function approximation explains the RBF Neural Network more stable and high efficiency.