Research On Clustering Algorithm Based On Spatiotemporal Trajectory Data And Its Application

In recent years,mobile Internet technology and big data science have developed rapidly.Electronic devices used in daily life have generated a large amount of user data,especially the spatio-temporal trajectory data with time and space characteristics.How to make full use of these spatio-temporal data,dig out potential valuable information from them,and make full use of them in urban planning,traffic,public security and other fields has become a new research topic and hot spot in the field of spatio-temporal data mining.Spatio-temporal trajectory data clustering is a branch of spatio-temporal data mining,aiming at discovering the movement rules of research objects and exploring their clustering patterns from massive spatio-temporal data.This paper mainly studies the clustering of spatio-temporal trajectory data,including the following three aspects.(1)Expansion algorithm based on temporal and spatial clustering of density peak.Although the existing spatio-temporal data clustering algorithm has good clustering effect in different scenarios,it cannot obtain reliable clustering results quickly due to the large number of parameters that need to be set and the great influence on the results.To solve this problem,this paper proposes an extended algorithm based on the rapid search of density peak spatio-temporal clustering.The algorithm expands the usage scenarios on the basis of the ST-CFSFDP algorithm,and is used to explore the popular spatio-temporal regions.On the one hand,according to the characteristics of the cluster center,the weight of each sample point is calculated and the cluster center is determined.On the other hand,considering the spatio-temporal continuity factor,constraints of time window and spatial distance are added in the process of sample point division,and the rule of sample point division is changed.Experimental results show that this algorithm can effectively explore the information of spatiotemporal hot regions.(2)According to the results obtained by the spatio-temporal clustering algorithm,a reference evaluation method for calculating spatio-temporal cluster sets is proposed by comprehensively considering the number of sample points in each cluster,the spatial location of the cluster and the time characteristics.Through this method,the spatio-temporal cluster sets with reliable information can be quickly selected.(3)Spatio-temporal trajectory clustering algorithm based on trajectory segmentation and boundary search.Due to the particularity of spatiotemporal data,the clustering results obtained by most common partition-based clustering algorithms can not distinguish the continuous and segmentation residence trajectory,resulting in errors in the final information of the trajectory residence region.In this paper,a spatio-temporal trajectory clustering algorithm based on trajectory segmentation and boundary search is proposed to explore the spatio-temporal trajectory residence area of a single research object.The algorithm uses the trajectory partitioning algorithm as the initial partitioning result of the clustering algorithm,and then obtains the trajectory partitioning result with clustering effect by searching the optimal boundary of each partition area.The experimental results show that the algorithm has a certain ability to recognize the area of space-time trajectory.