| Rapid construction of Urban rail in China proposes a challenge for traditional four-steps models of ridership estimation:On one hand,within denser rail network and rail station,multiplepossible stop-choices are faced by travelers to replace traditional shortest-distance-choices.In addition,rail demand prediction on using rail station is required as new rail lines and rail station are coming into use.This paper empirically examines rail demand on three typical cases:①new rail lines are using;②new traffic facilities are using around stations;③urban commercial center is moving.Firstly,basing on the whole life cycle of rail travel,connections between multiple dimensional representations of rail station and rail transit demand are analyzed.At the same time,the fluctuation of rail transit is tested by using time series clustering method to bulid a indicator system.Secondly,under the backgrand of big data within data crawling,data mining and data processing’s wider spread,framework of rail datais built.The statistical data and the crawling data are discriminatorily processed under such framework to estabilish a procedural rail date process.Moreover,based on multiple regression analysis,this study aims to predict rail demand on station level by using two forecase direct models:a rail ridership forecast model and a rail fluctuation forecast model.Finally,the methodology is applied to a real case study in NanJinwith the quality data of rail card.Predictions are finally evaluated by an a posteriori comparison with real data.This paper proposes that energy density on rail scale of service and spatial-configurational measures could be an important asset for urban rail demand forecast models at station level.The results highlight that by means of multiple regression analysis,it is sufficient to predict rail demand with common factors of multi-dimension on station level. |
