Estimation and pricing for substitutable products in choice-based revenue management

• Main Author: Bi, Yalin
• Other Authors: Currie, Christine ; Fliege, Joerg
• Published: University of Southampton 2018
• Subjects: 510
• Online Access: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.741727

-It has been proved that choice-based Revenue Management can result in significant increases in revenue in situations where a seller is pricing a set of substitutable products. This is particularly applicable to the transport industry and we present an example of train ticket sales. Estimating customer choice models is difficult, particularly in situations where the data file is incomplete. We use the Multinomial Logit (MNL) model to describe customer preferences, and a two-step algorithm to jointly estimate the parameters of this model and the customer arrival rate. A simple Markov Chain Monte Carlo (MCMC) method is also applied to update our belief of arrival rate and customer choice model. The dynamic programming model for the choice-based pricing problem suffers from the “curse of dimensionality ”. The computational time increases dramatically and makes it impossible to solve the problem with exact solutions. Approximate dynamic programming methods can be used to solve the problem. We propose a new approximation method that reduces the running time. The thesis will describe the complete methodology that we have implemented and provide some numerical results. As these are live sales systems, it is important that the system continues to earn revenues while the parameters are being estimated. A decision-making problem is needed to maintain a balance between the learning of customer preference (exploration) and earning (exploitation) in choice-based Revenue Management. In order to maximise the total revenue, the seller must decide whether to choose the current optimal price (exploitation) or to set prices that help to better estimate customer choice behaviour (exploration). We propose two pulling policies in a Multi-armed Bandit (MAB) experiment to balance the trade-off between exploration and exploitation.