119 / 2024-04-20 22:44:40

A Non-parametric Model for Price-setting Newsvendor problem under Demand Uncertainty

data driven,newsvendor,retail price,inventory

In this study, we explore a scenario where a retailer must set prices and order quantities without knowing full information of the demand distributions. The retailer tests only a few price points, observing demand outcomes for each. With this limited data, we develop a framework for considering non-parametric properties of the demand distribution and formulate two robust optimization approaches aimed at either maximizing minimum profit or minimizing potential regret. Our approach is data-driven, avoiding reliance on specific demand models, which helps prevent model mismatch inaccuracies in practice. Our results suggest that even with a small number of price testing, a significant portion of the maximum possible profit is achievable, both on average and in worst-case situations. We also show that our method outperforms traditional strategies like regression method and converges to the optimal with more price testing under mild conditions.