| In-depth study of the temporal and spatial characteristics of online car-hailing travel and the relationship between its influencing factors is conducive to optimizing the allocation of traffic resources,rationally planning urban land types,and making up for the deficiencies of the public transportation system,thereby alleviating traffic congestion and improving urban traffic conditions.However,due to the lack of publicly available online car-hailing data in China,research based on China’s online car-hailing data is quite limited.The paper is dataoriented,based on the large-scale spatiotemporal data of online car-hailing in Chengdu provided by Didi’s "Gaia Data Open Project",to carry out the analysis of the spatiotemporal characteristics of online car-hailing travel,and to explore the relationship between online carhailing passenger flow and various types of influencing factors.The main contents and conclusions of the study are as follows:(1)Processing and analysis of basic data.The paper selects the online car-hailing order data and influencing factors data provided by Didi’s "Gaia Data Open Plan" in the central city of Chengdu from November 1 to November 30,2016 as the basic data.After data preprocessing,format conversion and abnormal data cleaning of online car-hailing data,6,072,707 pieces of data were finally determined to be used in this study.The paper extracts the data of population density,land use mixture,road network density,public transportation proximity,average house price and 14 kinds of points of interest(POI)data as pre-selected influencing factor data,and visualizes and describes these data Statistical Analysis.(2)Analysis of spatiotemporal characteristics of online car-hailing demand.In terms of time distribution,the paper analyzes the weekly and daily variation characteristics of online car-hailing passenger flow,studies the characteristics of online car-hailing passenger flow during peak hours,and compares the difference in demand for online car-hailing on weekdays and weekends over time,and respectively.The trend of travel time of online car-hailing orders on weekdays and weekends was counted and compared.The paper analyzes and compares the spatial distribution of pick-up points and drop-off points for online car-hailing trips on weekdays and weekends,and finds that the passenger flow of online car-hailing has temporal and spatial regularity aggregation.Combined with the actual characteristics of residents’ online car-hailing trips,the paper visually analyzes the spatial distribution characteristics of various influencing factors in the study area.(3)Establish a spatiotemporal data analysis model.Through the screening of multicollinearity test,stepwise regression method and Global Moran’s I index,the paper identified restaurants,shopping,enterprises,transportation facilities,sports and leisure,business residence,government and administration,finance,science,education and culture,hotels,etc.Accommodation,population density,land use mixing degree,road network density,bus station density and average house price are 15 influencing factors as independent variables.For the dependent variable,three regression models were established: Ordinary Least Squares Linear Regression(OLS),Geographically Weighted Regression(GWR)and Geographically and Temporally Weighted Regression(GTWR)respectively.The model quantitatively analyzes the relationship between online car-hailing travel and its influencing factors.The paper expounds the basic theoretical basis of the three models,and compares and analyzes the characteristics,advantages and disadvantages of the three models.In addition,the paper introduces the key indicators of model comparison: Goodness of fit ,Adjusted ,Akaike Information Criteria()and revised ,namely .(4)Model results comparison and visualization analysis.The paper compared the results of the three models and found that the fitting performance of the GTWR model was better than that of the OLS model and the GWR model.After considering the spatiotemporal nonstationarity,the explanatory power of the GTWR model was significantly increased.In order to display and analyze the model results more intuitively and clearly,the paper conducts a visual analysis of the OLS model results,the spatial characteristics of the GWR model variable fitting coefficients,and the spatiotemporal characteristics of the GTWR model variable fitting coefficients.The research results show that various influencing factors are closely related to the demand for online car-hailing in the city.The influence of catering,shopping,enterprises,transportation facilities,science and education culture,hotel accommodation and road network density is the most significant.Among them,catering,shopping,transportation facilities,science and education culture,The impact of online car-hailing passenger flow and hotel accommodation has a positive promoting effect,while business and road network density have a negative inhibitory effect.The impact of finance on the passenger flow of online car-hailing is positively correlated on weekdays,but inhibits it on weekends;the impact of bus stop density on the passenger flow of online car-hailing is positively correlated,and the positive effect is more significant on weekends;The population density has a greater negative inhibitory effect on the passenger flow of online car-hailing;the positive promotion effect of the government and administration on the passenger flow of online car-hailing is greater on weekdays than on weekends.The other 4 variables,sports and leisure,business residence,land use mix and housing price,have little effect on the traffic of online car-hailing.The research is conducive to understanding the temporal and spatial rules of urban residents’ travel.Urban traffic management departments can use GTWR’s more accurate temporal and spatial models to formulate traffic demand management policies.Prediction can provide theoretical support for traffic management departments to allocate traffic resources for important influencing factors in newly developed areas,and provide valuable decisionmaking basis for urban planners in the planning and construction of new areas.At the same time,it can also help online car-hailing platforms in newly developed areas.Better promote the online car-hailing service,improve its operation management,and improve the quality of the online car-hailing service. |
PDF Full Text Request