一种基于简单遗传算法的K-Means改进算法  

Improved K-Means Algorithm Based on a Simple Genetic Algorithm

作　　者：尹鹏飞[1] 张晓丹[2] 

机构地区：[1]吉首大学教务处,湖南吉首416000 [2]吉首大学数计学院,湖南吉首416000

出　　处：《吉首大学学报（自然科学版）》2009年第6期43-45,53,共4页Journal of Jishou University(Natural Sciences Edition)

摘　　要：针对k-means算法对初始值敏感、易陷入局部极小值等缺点,结合遗传算法的思想,提出了一种基于遗传算法和k-means算法的混合聚类方法,为了测试该聚类算法的性能,用k-means算法和改进的算法进行了1组实验,并对2种算法的聚类结果进行比较,实验结果表明算法能够有效地解决聚类问题.K-means algorithm is sensitive to initial value, easy to fall into local minimum value. In response to these shortcomings, the idea of genetic algorithm is proposed based on genetic algorithm and k-means algorithm for hybrid clustering method. In order to test the performance of clustering algorithm,a set of experiments are conducted by using k-means algorithm and the improved algorithm, and the clustering results by the two algorithms are compared. It is showed that the clustering algorithm can effectively solve the clustering problem.

关 键 词：数据挖掘 聚类分析 遗传算法K—means算法 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论] TP183[自动化与计算机技术—计算机科学与技术]