基于空间划分的K-均值聚类分析优化算法

摘要

针对传统K-means算法随机选取初始聚类中心所导致的算法结果不稳定、易得出局部最优解的缺点,提出了空间划分初值聚类算法（CSPIC）.采用空间二分法得到初始聚类中心,从合理的初始聚类中心出发以实现更优的聚类效果,实验表明改进算法得出的聚类结果更具高效性和准确性.%Because of the problem that the random initialization of traditional K-means clustering algorithm leads to unstable results and local optimal convergence,clustering of space partition initial centroids（CSPIC） is presented.The initial centroids are obtained through space dichotomy,and then the high quality clustering result is produced on account of the rational initial centroids.The experiment demonstrates that the result of the improved algorithm is more efficient and accurate than that of the K-means algorithm.