Demand estimation and bundle price optimization : a data-driven approach

Abstract

The bundling of multi-products at a fixed price has become a popular marketing strategy and attracted many researchers’ attention. This dissertation investigates the bundle pricing problem with discrete choice models. Two demand estimation methods, Random-Coefficients Logit Model and Marginal Distribution Model, are carefully studied and implemented into a real data set in the fast food industry to exhibit their prediction ability. To solve the bundle pricing problem, we employ a framework called “Marginal Estimations + Price Optimization” developed by Yan et al., which is based on Marginal Distribution Model. The bundle price optimization is demonstrated by implementing this framework into the aforementioned data set. Besides, a demand forecasting method based on choice models is proposed and used to predict the market shares of new bundles in the context of bundle design.