Research And Implementation Of DBSCAN Algorithm Based On Spatial Clustering

With the development of human being in 20th century,the spatial data mining technology has been greatly developed.The DBSCAN(Density-Based Spatial Clustering of Applications with Noise)algorithm and St-DBSCAN(spatial-temporal Density-Based Spatial Clustering of Applications with Noise)algorithm has been widely used in cluster analysis of spatial data by data scientists,It has been used by scholars to cluster the GPS data of taxi in Kunming,and grasp the rules of residents' travel,so as to solve the difficult problem of Kunming residents'travel.However,the DBSCAN algorithm has the problem of long running time.For the St-DBSCAN algorithm,there are many problems in the algorithm,such as the long running time of the algorithm and the poor clustering results.Based on these two issues,this paper does related research work.Specific research work is as follows:(1)The improved DBSCAN algorithm is proposed according to the quadrant partition method.Firstly,based on the core point as the origin,the core point of neighboring points are divided into different quadrants,and then each quadrant area adjacent points and the number of neighboring points is deviated from the core point in each quadrant point center point as the representative point,complete cluster expansion operation.The improved DBSCAN algorithm,namely QD-DBSCAN(Quadrant-Division Density-Based Spatial Clustering of Applications with Noise)algorithm.(2)In this paper,firstly summarized the situation by tilting the distribution on the space object,there will be three kinds of data skew,namely core density tilt,boundary point density and noise density tilt.Then,aiming at the situation of each kind of density inclination,the corresponding improvement method is put forward based on the barycenter transfer,and describe the improved algorithm.(3)To detect he time performance and clustering effect of the improved St-DBSCAN algorithm and QD-DBSCAN algorithm,this paper implements the St-DBSCAN algorithm and improved QD-DBSCAN algorithm using Python,and the clustering results in the algorithm's time effect and the algorithm are compared,the result of the experiment is analyzed after the experiment.The experimental results show that the improved St-DBSCAN algorithm and QD-DBSCAN algorithm have a certain degree of improvement in time performance and clustering effect.