Research on Cluster Analysis of High Dimensional Space Based on Fuzzy Extension

Traditional spatial data are generally high dimensional features, and in the clustering of high dimensional data can be directly applied to data processing because of Dimension effect and the data sparseness problem. For CLIQUE algorithm, which usually have the problem such as prone to non-axis direction of overclustering, boundary judgment of fuzzy clustering and smoothing clustering. In this paper, a fuzzy clustering algorithm based on fuzzy extension of the highdimensional spatial data is proposed. This algorithm not only considers effect of adjacent grid of data points in the sparse grid, but also extending the sparse grid region to avoid smoothing clustering phenomenon occurs. What’s more, it can alleviate the problem of over-clustering and clustering fuzzy boundaries. Firstly, this passage gives a brief introduction to the characteristics of high dimensional spatial data as well as the Clustering methods. Based on the fuzzy extension a high-dimensional space data clustering analysis algorithms is put forward. The impact of the data samples around the point on the data points within the investigated grid is considered, sparse grid fuzzy extension, and at last, the problems of clustering fuzzy boundaries and to avoid excessive clustering produce meaningless clustering are solved.