系统聚类算法K-means 属于聚类分析中一种基本的划分方法，常采用误差平方和准则函数作为聚类准则，该算法在处理大数据集时是相对可伸缩且高效率的，同时具有潜在的数据并行性。但是这种算法依赖于初始值的选择以及数据的输入顺序；此外，当运用误差平方和准则函数测度聚类效果时，如果各簇的形状和大小差别很大，为使误差平方和 Jc 值达到最小有可能出现将大的聚类簇分割的现象。-system clustering algorithm K-means cluster analysis is a basic method is often used squared error criterion function as a cluster criteria, the algorithm in handling large data sets are relatively scalable, high-efficient and has the potential of data parallelism. However, this algorithm depends on the initial value of the options and data input sequence; In addition, when using square error of measurement function and the criteria clustering effect, if the cluster size and shape vary greatly, the square error for the Jc value and minimize the possibility of the emergence of large cluster segmentation clustering phenomenon.