Study On The Spatiotemporal Heterogeneity Of Ridership Determinants In Urban Rail Transit Stations

To alleviate negative effects brought by urban traffic pressure,such as air pollution,traffic jam and accidents,more and more cities take the construction and operation of urban rail transit as the first and best strategy.To achieve the service goal of increasing the attractiveness of urban rail transit,it is particularly critical to determine the key determinants of rail transit passenger travel and analyze the temporal and spatial evolution of its impact.Given that city-level rail ridership has non-stationary spatial-temporal stability.For example,morning and evening peak traffic in a certain area accounts for more than 50% of the daily traffic,which is a significant tidal phenomenon.Passenger travel is time sensitive and subject to history impact of passenger flows.Therefore,the general linear regression model used in previous studies that cannot explain the spatial heterogeneity of the influencing factors and the traditional Geographic Weighted Regression model(GWR)that cannot fully learn the key dimension of time have their own application limitations.In order to solve this problem,this paper used the big data network mining technology to target the one-month station ridership data of the Xi’an urban rail transit network in March 2018,and obtained the set of influencing factors for the station’s built environment.Then Geographic and Temporal Weighted Regression model(GTWR)was established to reveal the spatial-temporal impact of the built environment factors around the station on rail ridership.Firstly,this paper reviewed the research on the influencing factors of urban public transport ridership at home and abroad and the application achievements of the geographically weighted regression model.Secondly,the basic methods and model theories are elaborated in detail.Then,the mixed Poisson distribution model was used to process the subway smart card data,and the station classification according to the mixed land use was obtained,and then,the effective POI data was mined for the built environment to construct the set of impact factors of ridership.Through correlation analysis,the effective factor set was required for modeling.Finally,the comparative analysis was obtained by building traditional linear regression model,GWR model and GTWR model.Based on this point,the relationship between the rail ridership and the effective influencing factors were explored.The results showed that the goodness of fit of GTWR model is significantly improved,which can better explain the temporal and spatial impact of various influencing factors on rail station ridership than the GWR model and the global regression model.The main contributions include two points: First,based on GTWR model analyzes the relationship between Xi’an rail ridership and land use,and then discusses the impact of the built environment on the spatial and temporal distribution of rail ridership.Second,Effectiveness in terms of spatiotemporal impact of rail travel ridership is identified.