A Visual Analytics Approach To Studying Importance Of Urban Regions Based On Trajectory Data And Graph

During the process of urbanization,urban regions with different functions have been formed.Identifying the different functions of the regions and performing further visual analysis can effectively assist the urban managers to make reasonable policies.At the same time,the effective allocation of urban resources can meet the needs of people's lives.However,most of the methods for urban functional regions identification are mainly based on the types of land using,and cannot accurately reflect the functions of the regions where people's activities exist frequently.Human social activities are closely related to the distribution of urban functional areas.In this paper,we propose a method to identify the unban regions that has frequent people activities and discover different functions of these regions,based on trajectory data mining and POIs semantic analysis.Furthermore,an interactive visual analysis system is constructed to reflect the characteristics of human travelling.Different from the traditional region division methods that are mainly based on road network,we take the patterns of human activities into account,and identify the actual influence area of the functional regions by performing adaptive density clustering to the trajectory data and building Voronoi diagram based on the cluster centers.For each Voronoi cell,we build a regional functional document based on the POIs data in it,and mine the function of this region using LDA topic model.Through combining the analysis results with the visualization techniques such as map view,tree graph,streamgraph and circular heatmaps,we can explore the distribution of urban function regions from the overview to the details,and recognize the temporal moving patterns of the vehicles in each functional area.Based on the centrality measurements in graph theory,we evaluate the importance of each urban areas,and explore the characteristics of regional developments.The experimental results show that the proposed method can effectively identify the functions of urban areas,and help urban managers to make more rational decisions of urban planning.