基于旅客出行选择特征的高速铁路浮动定价策略研究

Research on High-speed Railway Floating Pricing Strategy Based on Passenger Travel Behavior Characteristics

基于旅客出行选择行为制定相应的定价策略,是提高铁路企业客票收益的关键。针对现行定价策略提出改进方案,首先构建考虑非时序特征的改进长短期记忆网络模型(LSTM),精准预测即将开售车次不同价格水平下的余票情况;其次,利用梯度提升决策树(GBDT)切割旅客出行选择特征空间,并依据特征进行价格分级;最后,构建以铁路企业收益最大为目标的定价模型,并采用逐次逼近算法求解。算例结果表明,改进LSTM模型预测余票标准化均方误差为0.053,比LSTM、RNN和GRU更优;改进的浮动定价策略与现行浮动定价政策相比,能够更好地反应运输市场需求,提高铁路部门单日收益。

The development of corresponding pricing strategies based on passenger travel choice behavior is the key to improve the ticket revenue of railway enterprises.An improvement scheme was proposed for the current pricing strategy.Firstly,an improved Long Short-Term Memory(LSTM)considering non temporal characteristics was used to accurately predict the remaining tickets of the upcoming trains at different price levels.Secondly,Gradient Boosting Decision Tree(GBDT)was used to cut the characteristic space of passenger travel choices and to classify the price according to the characteristics.Finally,a pricing model with the goal of maximizing the revenue of railway enterprises was constructed and solved by successive approximation algorithm.The results of the numerical experiments show that the standardized mean square error of residual forecast obtained by using the improved LSTM model can reach 0.053,lower than that of the ordinary LSTM,RNN and GRU.Compared with the current floating pricing policy,the new scheme proposed in this paper can better reflect the market demand,and increase the single day income.