传统的K-means聚类算法的研究与改进

The Traditional K-means Clustering Algorithm Research and Improvement

K-means算法是解决聚类问题的经典算法,在满足一定的条件情况下,聚类的结果比较好。但这种算法对初始聚类中心敏感,聚类结果随不同的初始输入而波动。针对这种缺陷,提出了一种新的基于数据样本分布选取初始聚类中心的算法。

The K-means clustering algorithm to solve the problem of the classical algorithm,under some conditions are met,the result of clustering is better.However,this algorithm is sensi-tive to the initial cluster centers,clustering results with different initial input and volatility.Ad-dress this shortcoming,a new distribution based on the data sample selection algorithm the initial cluster centers.