一种基于网格的K-Means聚类算法

A Kind of Grid-Based K-Means Clustering Algorithm

K-Means算法是聚类方法中常用的一种划分方法。随着数据量的增加,K-Means算法的局限性日益突出。基于网格划分的思想,提出了一种基于网格的K-Means聚类算法,该算法使用了网格技术在一定程度上去除了孤立点和噪声数据,减少了原始K-Means算法将大的聚类分开的可能。实验表明,该算法能处理任意形状和大小的聚类,对孤立点和噪声数据也能很好地识别,并且在去除孤立点和噪声数据方面可以达到较好的精度。

K-means that is a kind of partition method often be used in the clustering.This paper presents the Grid-based K-means Clustering Algorithm,which voids the phenomenon of the local clustering result.The Algorithm reduces the probability that a cluster will be divided into some clusters by the use of the error square and rule function.The algorithm deals with outliers by the technique of grid-based.The experiment results show that it can discover outliers or noises effectively and get good cluster quality.