Research of Hierarchical Clustering Based on Dynamic Granular Computing

Hierarchical clustering algorithm of cluster analysis method can detect data in different granularity levels, and the clustering process is suitable for most of practical problems, but complexity of the merger and split conditions limits its application. This paper proposed a new hierarchical clustering algorithm basis on dynamic fuzzy granular computing combining with the particle swarm optimization, the condition of merger and split were new expressed and optimized. The experimental results show that the new algorithm is effective.