一种面向数据密集型计算环境的聚类算法

A Clustering Algorithm for Data-Intensive Computing Environments

针对数据密集型计算环境下数据具有海量、分布、异构、高速变化等特点,分析传统的基于密度的分布式聚类(Density Base Distributed Clustering,DBDC)算法,借助MapReduce编程模型,提出一种新的分布式聚类算法,采用局部和全局的方式处理海量、异构数据,解决具有以上特点的数据密集型计算环境下数据的分析挖掘问题。得出算法的复杂度为O((nlog2n)/p),实验验证在数据量与节点数变化时算法具有较高的稳定性和可伸缩性,与原算法对比该算法具有较高的准确度。

DBDC(Density Base Distributed Clustering) is a distributed clustering algorithm which is based on the density clustering DBSCAN algorithm.In this paper,we research the DBDC algorithm,and propose a new IDBDC algorithm for the data-intensive computing environments based on the MapReduce model,and discuss the complexity of the algorithm.Experiments verify the feasibility and effectiveness of the algorithm.