[1]张忠林,曹志宇,李元韬..基于加权欧式距离的k_means算法研究[J].郑州大学学报(工学版),2010,31(01):92.[doi:10.3969/j.issn.1671-6833.2010.01.022]
　ZHANG Zhonglin,Cao Zhiyu,Li Yuantao.Research on k_means algorithm based on weighted Euclidean distance[J].Journal of Zhengzhou University (Engineering Science),2010,31(01):92.[doi:10.3969/j.issn.1671-6833.2010.01.022]

点击复制

基于加权欧式距离的k_means算法研究()

分享到：

《郑州大学学报(工学版)》[ISSN:1671-6833/CN:41-1339/T]

卷:

31卷

期数:

2010年01期

页码:

92

栏目:

出版日期:

2010-01-30

文章信息/Info

Title:

Research on k_means algorithm based on weighted Euclidean distance

作者:

张忠林; 曹志宇; 李元韬.

兰州交通大学电子与信息工程学院,甘肃,兰州,730070

Author(s):

ZHANG Zhonglin; Cao Zhiyu; Li Yuantao

DOI:

10.3969/j.issn.1671-6833.2010.01.022

文献标志码:

A

摘要:

传统的k_means算法将欧式距离作为最常用的距离度量方法.针对基于欧式距离计算样本点与类间相似度的不足,用"相对距离"代替"绝对距离"可以更好地反映样本的实际分布,提出一种在领域知识未知的情况下基于加权欧式距离的k_means算法.针对公共数据库UCI里的数据实验表明改进后的算法能产生质量较高的聚类结果.

Abstract:

Traditional k_means algorithms use Euclidean distance as the most commonly used distance measurement method. Aiming at the shortcomings of calculating the similarity between sample points and classes based on Euclidean distance, replacing "absolute distance" with "relative distance" can better reflect the actual distribution of samples, and a k_means algorithm based on weighted Euclidean distance is proposed when the domain knowledge is unknown. Experiments on the data in the public database UCI show that the improved algorithm can produce high-quality clustering results.

常用功能