User Behavior Analysis And Visualization Based On E-bike Data

In recent years,electric bike(e-bike)industry has developed rapidly,greatly improving the transportation network and increasing the efficiency of traveler.However,it also brings many traffic and social problems,such as disturbing traffic order,causing traffic accidents,and e-bike being stolen frequently.In order to solve the problems mentioned above,management department needs to comprehensively understand the behavior of e-bikes and govern them to make them better serve the social economic development.In recent years,some cities in China deployed intelligent tracking system for e-bikes.The system has collected huge number of realtime e-bike location data,based on which we can mine valuable information for e-bike management with big data technology.In this thesis,we carry out systematic research work based on e-bike data from some city in China,not only studying the behavior of e-bike users,but also designing a visualization system with colorful functions.Our contributions are summarized as follows:We first mine stay points and moving trajectories from e-bike data.Based on stay points,an algorithm is proposed to automatically detect e-bike user’s home location.In order to study user behavior to stay points,we use the place preference matrix to model their routine behavior.on the basis of routine behavior,this paper proposes a user similarity calculation approach.Then,we design a user clustering model based on Kmeans,and determine the optimal clusters with DB Index.Experiment shows that the model can effectively distinguish the behavior of users,and help people understand the distribution of their habits and occupation.To estimate the ride speed with e-bike moving trajectories,this paper designs a map match-ing algorithm based on path constraint.The algorithm determines travel path using the spatial relation among data to improve the matching accuracy,and kd tree,A*algorithm and dynamic programming method are synthetically used to ensure high computational efficiency.On the basis of data analysis results,we also builds a Web visualization system to facilitate the intuitive analysis and understanding of e-bike data.The system supports the function of electric bicycle monitoring,electric bicycle mobility analysis and user behavior analysis,making it possible for the practical application of our research work.