Research On Censored Railway Passenger Transportation Demand Estimation Based On Non-parametric Choice Model

Accurately grasping passenger transport demand is the premise for railway passenger departments to do a good job in marketing.Historical ticket sale data comprehensively records passenger ticket purchasing behavior,contains abundant passenger demand information,and is the most common data source for passenger flow analysis and prediction.However,due to the structural deficiencies of the capacity,there will be a shortage of tickets on certain dates,certain time periods,certain trains,and certain ODs.At this time,some passenger requests are denied,and the historical ticket sale data obtained in the ticket system does not represent the actual passenger demand,which affect the passenger flow analysis and prediction accuracy,and influence railway passenger transport marketing decision science and rationality.Therefore,how to accurately restore the real ticket purchasing demand of passengers from the historical ticket sale data is an urgent problem to be solved by the railway passenger transport department,which is also the problem that this paper tries to solve.This paper first analyzes the railway passenger ticket purchase process,which includes the passenger arrival process and ticket purchase decision process,and analyze passengers’ decision-making behavior under ticket quantity censored conditions,and determines the determining method of product censored considering the sold and remaining ticket data.Secondly,a ticket demand estimated model based on historical ticket sale data,including passenger arrival model and ticket selection model,is constructed,and the solution steps of the model based on EM algorithm are given.Then,the Monte Carlo method are used to evaluate the accuracy of the model,and the stability of the model in multiple scenarios was discussed by designing different experiments.Finally,an empirical analysis based on the actual operational data is conducted to verify the performance of the model in practical application.The research results of this paper provide a new idea for railway ticket demand estimation,help railway passenger transport departments to better grasp the real passenger flow situation,and provide support for railway passenger transport departments to make marketing decisions.19 figures,17 tables,56 references.