考虑对象方向关系的密度聚类算法

A clustering algorithm based on density and direction

摘要: 聚类分析是数据挖掘的一个重要研究方向.为了在大规模空间数据库中发现任意形状的聚类,Martin Ester等提出基于密度的聚类算法DBSCAN.针对DBSCAN处理聚类边界对象的不足,提出了聚类时考虑对象方向关系的改进算法,实验表明,改进算法在不改变时间、空间复杂度的情况下能得到更好的聚类结果.

Abstract: Cluster analysis is a primary method for data mining.Martin Ester introduced a algorithm DBSCAN relying on a density-based notion of clusters which was designed to discover clusters of arbitrary shape on large spatial databases. An improved DBSCAN algorithm is presented.The results of our experiments demonstrate that our algorithm is more effective without changing efficiency.