Visual Analysis And System Development Of Sparse Traffic Trajectory Data

With the rapid development of urbanization,city life is becoming more diversified and complex,and people's daily travel activities produce a large number of spatial trajectories.Trajectory data containing spatial,temporal and semantic,is very helpful for the government in urban planning and the understanding of public behaviors.Trajectory analysis now becomes a hot topic in many research fields.In this paper,we focus on visually discovering and analyzing users' movement patterns from driving and public transport data and propose two visual analysis systems for sparse traffic trajectory data.The sparse traffic transportation cell data is used to study movement patterns of driving.We first explore the macro-traffic patterns,and design a visual query model based on the algorithm of fake plate vehicle detection.We then use the spatial bubble chart,CirFlow chart and travel time distribution chart to analyze the spatial and temporal distribution of query results in the global map view,single cell view,and cell pairs view.We finally interactively analyze the flow distribution of traffic cells,the process of fake plate vehicle detection and so on.Case studies demonstrate the effectiveness of our system in finding movement patterns of fake plate vehicles.Movement patterns of public transport data are described based on the trip data of bike share systems.We construct a spatial-temporal-user tensor factorization model,and apply Non-negative Tensor Factorization(NTF)on it to decompose the tensor into the basic mobility pattern tensors.The time,space and user dimensions are visualized by the time chart,heat map and age meter graph,respectively.Users can interactively explore the time,space and age distribution in different basic mobility pattern tensors.We apply our approach to New York and Chicago bike share data,cases show the usability of our system in exploring users' behavior patterns.