Study On The Difference Of Spatial Demand Between Traditional Taxi And E-hailing Taxi And Its Influencing Factors

Traditional taxis and e-hailing taxis are important means of transportation in the urban transportation system.The analysis of the spatial distribution of taxi demand is critical because it can help us understand the preference of passengers and drivers of the taxi.The purpose of this paper is to analyze the difference in the spatial distribution of traditional taxi and e-hailing taxi demand.We study factors that influence traditional taxis and e-hailing taxis demand by using trip data of traditional taxis and e-hailing taxis as well as land use,social economy and transport-related data in New York City in 2017.Taxi zones defined by the New York City Taxi and Limousine Commission(NYCTLC)is used as the basic study unit.The main contents of this paper are as follows: 1)Summarize the relevant literatures that affect the demand of traditional taxis and e-hailing taxis.2)Describe the data used in this study as well as the data analysis process.3)Introduce the statistical models used in this paper,including spatial autoregressive model and fractional regression model.4)Analyze factors that influence ridership of traditional taxis and e-hailing taxis by constructing spatial autoregressive models for traditional taxi and e-hailing taxi respectively.5)To understand the difference between traditional and e-hailing taxis,we construct a fractional response model to study factors that influence the proportion of different zones.The results show that there are some differences in the space-time distribution of trips between the traditional taxi and e-hailing taxi.Results of Moran's I show that there is strong spatial dependence in taxi ridership.If this spatial dependence is ignored,it will lead to biased parameter estimation and unreliable statistical inference.The spatial autoregressive model can produce a more accurate explanation of the external factors that affect the demand of the traditional taxi and e-hailing taxi respectively by controlling the spatial dependence.In the model of traditional taxis,park area,population density,employment density,the proportion of population over 65 years old,the proportion of household with income higher than 200,000 dollars,and existence of airport are positively related to trips traditional taxi trips,while The proportion of household with income lower than 50,000 dollars and subway stations density are negatively related to traditional taxis trips.In the model of e-hailing taxi,population density,employment density,the proportion of population between 18 and 34 years old,bike line density,and existence of airport are positively related to e-hailing taxi trips,while park area and parking density are negatively related to e-hailing taxi trips.The spatial distribution of the proportion of e-hailing taxi trips shows that the farther away from the Manhattan area,the higher the proportion of e-hailing taxi trips.The results of fractional regression model show that residential area,manufacturing area,the proportion of population between 35 and64 years old,the proportion of household with income less than 50,000 dollars,and distance from the city center are positively related to the proportion of e-hailing taxi trips,while park area,population density,bus station density,and existence of airport are negatively related to the proportion of e-hailing taxi trips.