关键词: 数据挖掘, 聚类, 密度函数, 密度可达, 划分方法

Abstract: A clustering algorithm based on density and partitioning method is presented according to the analysis of the strengths and weaknesses of traditional clustering algorithms.The algorithm can automatically locate the dense centers of clusters,and determine initial partitions of the clusters.On the basis of initial partitions of the clusters,density reachable clusters of data objects are found out by using partitioning method,and the final clusters are produced.The experimental results demonstrate that the algorithm can handle clusters of arbitrary shapes and sizes,minimize the influences of noise and deviation of data objects,and locate the outliers.At the same time,the algorithm can minimize the dependency of input numbers on specialist knowledge.

Key words: data mining, clustering, density function, density reachable, partitioning method