Clustering validity index is used to evaluating the clustering result yielded by the fuzzy clustering algorithm. In this paper, a new cluster validity evaluation approach is proposed to determine the optimal fuzzy c-partition produced by the fuzzy c-means algorithm. The proposed evaluation method introduces random sampling method to calculate distribution density. The results of experiment performed on three various data sets indicate that the proposed index is effectiveness comparing with the old validity index based on density. Especially, for spatial data clustering validity evaluation, random sampling can reduce the computation complexity.
展开▼
