Exploring Spatial-Temporal Characteristics And Influencing Factors Of Online Car-Hailing Ridership With Multi-Source Data

With the development of big data technology,the use of massive data to study traffic problems has attracted more and more attention from researchers in the field of transportation.As a booking-based transportation service relying on the Internet platform,the emerging online car-hailing has generated a lot of data over the past decade.The rapid development of the online car-hailing also resulted in many problems.For example,the imbalanced spatial-temporal distribution of online carhailing passenger flow,has led to various problems such as insufficient e-hailing cars in peak areas during rush hours.Meanwhile,traditional methods(survey and statistics)to obtain the influencing factors of passenger flow are inefficient and costly.Therefore,the research aims to analyze the determinants of e-hailing passenger flow based on Point of Interest(POI)data,which provides methodological guidance to solve the problem of imbalanced supply and demand.POI data,map data and the data of e-hailing ridership provided by Didi are combined to analyze the spatial and temporal distribution characteristics of e-hailing ridership.The relationships between e-hailing ridership and all kinds of POIs are established,and the results are visualized.The main research work are as follows,(1)In terms of the temporal and spatial distribution of online car-hailing passenger flow,the analysis is conducted from three time domains i.e.morning and evening peaks,24 hours,weekdays and weekends.Meanwhile,based on the reclassifies POI,the spatial distribution of e-hailing ridership are presented and compared between weekdays and weekends.(2)Using the number of POI types in the study area as independent variables and e-hailing ridership as dependent variables,the least squares linear regression model,geographic weighted regression,and geographically and temporally weighted regression modeling are used for modeling and comparative analysis.(3)A variety of visualization methods are used to analyze and present the modeling results,analyze the influence mechanism of each parameter.Based on the current e-hailing operation status,countermeasures and suggestions for the future operational management of e-hailing are put forward.The results show the passenger flow of e-hailing has non-stationarity in time and space.And POI data can better characterize the contributing factors to e-hailing ridership and reveal the influencing mechanism.Geographically and temporally weighted regression model can fit the passenger flow of e-hailing best,which can be used to predict the spatial-temporal distribution of the passenger flow of e-hailing in a long term.The results of the study can be useful guidance for the regulation of the e-hailing industry,the adjustment of imbalance between the supply and demand of ehailing vehicles and the regional and time-based dynamic pricing of e hailing vehicles.