基于复合形的K-means优化聚类算法研究  

K-means Optimal Clustering Algorithm Based on Complex

作　　者：赵凯[1] 李声晋[1] 赵锋[2] 

机构地区：[1]西北工业大学机电学院,西安710072 [2]西北工业大学航空学院,西安710072

出　　处：《郑州大学学报（理学版）》2008年第4期44-47,共4页Journal of Zhengzhou University:Natural Science Edition

摘　　要：为了克服K-means算法受初始点影响大、结果稳定性差的不足,提出了一种新的K-means优化聚类算法.介绍了复合形法的基本原理并将其做了一定修改以适用于K-means优化聚类,推导了一系列用于计算的公式,给出了具体的实现步骤与方法.通过算例说明,与其他几种方法相比,该方法结果稳定,计算效率较高,有着很好的推广应用前景.A novel K-means algorithm is presented to overcome the weakness of the result stabilization and the effect of the initial point. First,the main principle of the complex method is introduced and an improved one is used to K-means optimal clustering,a series of formulation is educed in this paper and the realization process of the algorithm is given later. Compared with other algorithms, this method has a stable result,an efficient calculation, so it has very good application and dissemination prospect.

关 键 词：复合形 数据挖掘 K-MEANS聚类 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]