A comparative analysis of demand estimation approaches in yield management

The objective of this paper is to provide a comparative analysis of demand estimation approaches in the context of yield management. We propose different theoretical modeling approaches for demand estimation. Demand is assumed to be function of price, time and inventory remaining. The proposed models are – intrinsically linear model, nonhomogeneous Poisson demand process (Leemis 1991), and a neural network approach. These are compared with three benchmark models: Swami and Khairnar (2003), Tretheway and Weinberg (1991) model, and price‐based model. Four representative demand data sets from the car rental business are used (Wang 2001). Our empirical analysis results show that the proposed intrinsically linear model and neural network approach provide good estimates of demand.