Research On Analysis Method Of Urban Road's Cor-Relation With Taxi Trajectory Data

Urban computing rooted in urban big data is becoming an important way to solve problems like traffic congestion and urban planning.It is becoming more important to use vehicle trajectory data in urban computing research because it contains abundant traffic information and has a natural connection with the urban road network.Knowing well about the correlation between urban roads is of great significance in solving traffic jams and optimizing road planning.Because vehicles' moving data between roads prob-ably contains the correlation of roads,this paper studied the method of analyzing road correlation from trajectory data.The main content is as follows:(1)Proposed the "road to vector" method:"word to vector" is an important method in natural language processing.It can generate similarity from original incommensurable words by generating vectors of words,which laid the foundation of many applications.This article proposed the concept of "road to vector" by comparing the road to the word in the method,and the trajectory to the sentence,and applies the word embedding tech-nique to generate vectors of roads,which were used to calculate road correlation to sup-port analysis of urban road.(2)A visual analysis system of road correlation:Based on the "road to vector" meth-od,a visual analysis system of road correlation was designed and implemented,includ-ing map view,projection view,timeline view and other associated views.It allowed us-ers to analyze trajectory data from different scales and perspectives,and explore the road correlation interactively.(3)Road correlation analysis and verification:Taking the taxi trajectory data of Wen-zhou as an example,this paper used the method to analyze the road correlation and ob-tained some characteristics of urban traffic.By explaining the results,the validity of the method was verified.