全局K-均值聚类算法需要随机选取初始的聚类中心，本文基于K中心点算法的思想，将其作为全局K-均值聚类算法的初始聚类中心，并对全局K-均值聚类算法进行改进。依托人工模拟数据和学习库中的数据分析，对比两种算法的性能，得出改进算法聚类时间短，鲁棒性强的结论。

The global K - means clustering algorithm needs to randomly select the initial clustering center. In this paper, the K -medoids is used as the initial clustering center, and the global K - means clustering algorithm is improved. With the data analysis in the database, the performances of the two algorithms are compared. It concludes that the clustering time of the improved algorithm is short and its robustness is strong .