Research On High Speed Railway Revenue Management Based On Passenger Choice Behavior

In recent years,with vigorous construction and operation of high-speed railways in China,railway transportation capacity has been significantly improved,which also drives economic development of regions along routes.At the same time,with reform of China's Railway marketization,previous operation and management experience has been unable to adapt to today's fierce external competition.It is necessary to discuss revenue management from perspective of passengers,using market-oriented means to adjust supply and demand relationship,improve service and enhance comprehensive competitiveness of high-speed railways to gain greater market share.This article starts from applicability conditions for implementing revenue management,and compares foreign railways and aviation with high-speed railways in China,then discusses its feasibility.Revenue management is introduced from three aspects of passenger classification,dynamic pricing and inventory control.After compares advantages and disadvantages of several customer classification algorithms,a CART(classification and regression tree)algorithm is selected for passenger classification;then analyzes main problems of China's high-speed rail fare system,and leads to direction of implementing dynamic pricing theory.Combining with current ticket allocation method of China's railways,key and difficulties of implementing seat control are explained.Finally,relationship between fares,stocks,and demand in railway revenue management is analyzed,and a joint study on dynamic pricing and seat control issues is proposed.This paper researches from these four steps through literature retrieval,data mining,model construction,and result analysis:(1)Through reading literature,this paper analyzes background and significance of high-speed railway revenue management researches,summarize research status at home and abroad from these three aspects: Dynamic pricing,Seat control and Joint research.Analyze feasibility of implementing revenue management in China's railways,then compare air transportation with railway from these aspects: network scale,characteristics and ticketing methods,finally puts out key to implementation of revenue management in China's high-speed railways.(2)In passenger classification model,in order to overcome shortcomings of traditional decision tree model,two ensemble learning algorithms based on CART are introduced to enhance generalization ability of classification and regression model.Random forest model is used to classify passengers into discount ticket passengers and full-price ticket passengers.Based on the classification results,demand relationship between two types passengers is obtained.To find out crucial feature in classification,a feature importance ranking model is applied.Gradient boosting tree model is use to forecast optimal discount rate,using examples to verify model respectively.(3)Combining classification and prediction results,a high-speed railway ticket allocation model based on discounted tickets is proposed,and particle swarm optimization algorithm is designed to solve INLP.Based on model results,a comparison scheme is designed: Discount demand is greater than demand for full-price tickets,which verify effectiveness of original scheme model.Final result shows that ticket allocation model based on discount tickets can obviously attract low-demand segment passenger flow,and bring certain ticket revenue while ensuring attendance rate.