Hotspots Detection And Dynamic Analysis From Trajectory Data Based On Spatio-Temporal Data Field Clustering And Complex Network

City hotspots refer to the area where residents visit frequently,and Social hot events happen,which reflects the people travel patterns and effectiveness of urban planning.Extracting and analyzing the spatio-temporal characteristics of city hotspots can help residents to choose the destination and avoid the rush before traveling.Besides that,the study of hotspots can warn the traffic related department of high-traffic locations early,so that they may allocate police and make the emergency plan in advance.For city planning department,the detection of hotspots can be used to evaluate the effect of regional planning and prepare perfect city planning scheme.As a kind of important public transportation,taxi trajectory data contain abundant information about urban functions and citizen activities owing to its long service time,wide coverage of city area and freedom of its motion.The pick-up and drop-off points in the taxi trajectory data depict the spatio-temporal event of passenger or driver behavior patterns.Thus,we regard the area where the pick-up and drop-off points cluster as the city hotspots area.In this paper,we introduce a data field-based cluster analysis technique to the pick-up and drop-off points of taxi trajectory data,and present an improved method for spatio-temporal clustering by introducing the time weight,which has been normalized to estimate the potential value.To further analyze the interaction among the city hotspots,we apply the complex network theory to the cluster result,and build a city spatial network based the hotspots and region in Wuhan.The main research work of this paper is as follows:(I)data filed based clustering methods with the temporal extensionsThe pick-up and drop-off points are irregular,and the interaction among them would decline over time.Aiming at this problem,the data filed theory has the advantage about identifying clusters with different shape and uneven density distribution.But the traditional data field theory cannot satisfy the demand of spatio-temporal cluster.In the paper,we improve the data field potential function by introducing the time weight,which has been normalized to estimate the potential value and put the averaging difference value of potential as the standard of measurement of similarity among points.Thus,in the light of the new potential function,short distance and short time difference play a powerful role,that is,the region full of trajectory points has a higher potential value,while the region with thin trajectory points has a lower potential value.(II)Hotspots detection and analysis based on spatio-temporal data field clusteringTo find the interesting and hidden knowledge from taxi spatio-temporal trajectory data,we conduct the experiment on the pick-up and drop-off points to detect the city hotspots in Wuhan.The data in holiday,workday and weekend are clustered respectively,and we find the different spatial distribution of hotspots from the result.Moreover,we analyze the driving factors of hotspots area from the perspective of social event,the policy of urban planning and the difference of residents behavior.(III)The analysis of intra city networkBased on summarizing the classical theory and main statistical indexes of complex networks,the regional interaction network between the administrative divisions and the hotspots area is built respectively.Especially,the edge is the resident trips between the areas.Then we apply the statistical indicators of complex network to the intra city network.The interaction among the areas would be qualified,and the importance and influence of the areas would be depicted.