一种改进的k-中心聚类算法研究

An Research on Modified k- medoids Data Cluster Algorithm

聚类算法的好坏直接影响聚类的效果。文章讨论了经典的k-中心聚类算珐,说明了它存在不能很好地处理大数据集的问题,提出了一种基于加权改进的k-中心聚类算法,克服了k-中心聚类算法的一些缺点,并从理论上分析了该算祛的复杂度。实验证明,用该方法实现的数据聚类与传统的基于中心的方法相比较,能有效提高数据聚类效果。

The memod of data clustering will influence the effect of clustering directly.The algorithm of k-medoids is discussed,the shortages of this algorithm such as it call not deal with the problem of large data sets.A modified k-medoids clustering algorithm based on weights is put forward,it changes the shortcomings of k-medoids. Its complexity is analyzed from theoretical .The experimems show that compared with traditional method based on mediods,the modified data clustering algorithm can improve the efficiency of data clustering.