Research On Uncertain Spatial Data Clustering Method

Cluster analysis is an important branch of the data mining, and it has important applications in the analysis of internal data structures. Cluster analysis is an unsupervised learning of the classification process, which aimed unmarked sample data automatically into the meaningful groups or clusters through the algorithm. When data is assigned to the same group, the internal structure of all sample data similarity is high, or it not. Cluster analysis technique not only can find the distribution of sample data quickly, but also can reveal the distribution of sample space.For the clustering techniques, this paper mainly studied the following sections:First, UK-means algorithm is very sensitive to outliers in dealing with uncertain data,which must be acquired of the probability density or distribution function of uncertain data in advance. However, it is often difficult to obtain in practice. For the shortage of UK-means in dealing with uncertainty measurement data, this paper firstly proposed to a new algorithm :U-PAM. the U-PAM algorithm can analyze the clustering result by combining with the CH validity index to determine the optimal clustering number. Experimental results show that:the proposed algorithm can give effective clustering result obviously.Second, this paper proposes UM-DBSCAN clustering algorithm, which combines the DBSCAN clustering algorithms and intervals. it aims to transform the uncertain data to the center data. it can find Arbitrary shape of the sets.Finally, this paper introduce to the clustering algorithms about massive uncertain data: UM-PAM, UMC-DBSCAN. they aim to complete to cluster for the missive uncertain data.