Abstract: Aiming at the problem of the clustering analysis on massive data for traditional clustering algorithm, this paper proposes a K-means clustering ensemble algorithm based on MapReduce. It generates component clustering results with different number of cluster by the K-means algorithm, improves co-association matrix, and gets a final result automatically via the number of times sample pair co-occurred. Experimental results show that this algorithm can effectively improve the quality of clustering, and has good scalability, fits to clustering analysis on massive data.

Key words: massive dada, clustering, MapReduce framework, K-means algorithm, co-association matrix, clustering ensemble

摘要： 针对传统聚类算法难以高效进行海量数据聚类分析的问题，提出一种基于MapReduce框架的K-means聚类集成算法。利用K-means算法生成不同聚簇数目的基聚类结果，改进共协关系矩阵，依据数据点对出现次数进行集成，自动得出最终聚类结果。实验结果表明，该算法能够有效地改善聚类质量，具有良好的扩展性，适用于海量数据的聚类分析。

关键词: 海量数据, 聚类, MapReduce框架, K-means算法, 共协关系矩阵, 聚类集成

CLC Number: 

• TP311