一种改进的基于密度的聚类算法

A modified density-based clustering algorithm

基于密度的聚类是聚类算法中的一种,其主要优点是可以发现任意形状的簇,但处理大数据集时效果不佳,为此提出了一种改进的算法M-DBSCAN,保留了基于密度聚类算法的优点,同时克服了以往算法不能处理大数据集的缺点。实验结果证明,M-DBSCAN聚类算法在聚类质量及速度上都比原DBSCAN有较大提高。

Density-based clustering analysis is a kind of clustering analysis method that can discover clusters with arbitrary shape. However, clustering analysis cannot deal with large database efficiently.In this paper, a modified algorithm M-DBSCAN is presented. It keeps the good feature, and it also congure the fault that the density-based clustering cannot deal with large database. The test proves that the modified algorithm is improved in the quality and rate.