传统的K-medoids聚类算法的聚类结果随初始中心点的

不同而波动，且计算复杂度较高不适宜处理大规模数据集；

快速K-medoids聚类算法通过选择合适的初始聚类中心改进

了传统K-medoids聚类算法，但是快速K-medoids聚类算法

的初始聚类中心有可能位于同一类簇。为了克服传统的K-

medoids聚类算法和快速K-medoids聚类算法的缺陷，提出

一种基于粒计算的K-medoids聚类算法。-Traditional clustering K-medoids clustering algorithm with the initial centers

Different swings, and a high degree of computational complexity inappropriate handling large data sets

Fast K-medoids clustering algorithm by selecting appropriate initial cluster centers to improve

The traditional K-medoids clustering algorithm, but fast clustering algorithm K-medoids

The initial cluster centers may be located in the same class clusters. In order to overcome the traditional K-

Medoids defect clustering algorithm and fast K-medoids clustering algorithm,

K-medoids one kind of clustering algorithm based on granular computing.