经典划分聚类分析方法及算例

Summary of Two Classic Partitioning Cluster Analysis

聚类分析方法在大数据时代是数据挖掘(Date Mining)、科学研究和机器学习(Machine Learning)的基础性工具。本文在收集、整理已有研究成果的基础上,对K-均值和K-中心点这两种经典划分式聚类分析方法的原理、优缺点及适用范围进行了重点阐述,并通过算例对比说明孤立点对两种经典算法的不同影响,对科研工作者更高效和便捷地寻求适用于自己研究领域的聚类分析方法、取得科学有效的研究成果具有重要意义。

Clustering Analysis is the basic tool of data mining,scientific research and machine learning in big data era.Based on summarizing the previous studies,the theories,pros and cons of the two classic clustering analysis methods were explained,which were called K-means and K-medoids.Through the numerical examples of isolated points,the difference between the two methods was indicated.This paper had significance to help researchers choose suitable cluster analysis method more efficiently and conveniently.