Research On Uncertain Data Clustering In Obstacle Space

With the development of information technology,uncertain data clustering is playing an increasingly important role in economic,military,and telecommunication internetwork.However,obstacles such as rivers,lakes,and valleys often exist in the real world.Therefore,the study of u ncertain data clustering in obstacle space has high practical value,making the results of cluster analysis more practical.In order to solve the problem that traditional clustering cannot effectively deal with the clustering of uncertain data in the obstacle space,a method to quickly calculate the obstacle bypass distance and the obstacle connection distance between two data points is proposed,and then based on the DBSCAN algorithm,The OEU-DBSCAN algorithm is being proposed.The OEU-DBSCAN algorithm uses obstacle distance as a measure of similarity between data,re-defines core points as obstacle probability core points according to the probability of data existence,and clusters data by obstacle probability core points.Subsequently,the FPA-EU-DBSCAN algorithm was proposed,and the efficiency of the algorithm was further improved by filtering the obstacle distance calculation.Experimental analysis shows that the proposed algorithm has higher efficiency and accuracy when processing uncertain data in the obstacle space.For improving the obstacle space does not determine the efficiency of data clustering algorithm,based on obstacle space Range uncertain data clustering algorithm of Tree,Tree Tree algorithm of data to establish the Range of uncertainty,and vertex Tree Tree to establish the Range of the rectangular area data query and obstacle vertex to determine the visibility of data in a rectangular area,in order to reduce the frequency of region query in the algorithm,and algorithm adopts the method of selecting representative point,under the condition of almost no loss of accuracy of clustering to further improve the efficiency of clustering.When the obstacle is large,the obstacle passes through the rectangular area and no obstacle vertex is in the rectangular area,thus reducing the visibility accuracy of data in the rectangular area.To solve this problem,a Range T-ROUDBSCAN algorithm is proposed,which takes equal points on the edge of obstacles,sets up a Range Tree for the equal points of the obstacle vertices and edges,and determines the visibility of data in the rectangular region by asking Range Tree for the equal points of the obstacle vertices and edges in the rectangular region.Experiments show that the Range T-ROUDBSCAN algorithm has high efficiency and clustering accuracy.