Algorithms for bi-level optimization in retail: pricing, assortment and demand coordination
This dissertation focuses on extensions of product line optimization modeling. Retail stores are faced with many decisions to optimize the process of satisfying customer demand. Thanks to current technologies, they can make use of diverse, relevant information in their decision making. Specifically,...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis doctoral |
| Estado: | Versión publicada |
| Fecha de publicación: | 2021 |
| País: | Chile |
| OAI Identifier: | oai:repositorio.anid.cl:10533/253176 |
| Acceso en línea: | https://hdl.handle.net/10533/253176 |
| Access Level: | acceso abierto |
| Palabra clave: | Ingeniería y Tecnología Otras Ingenierías y Tecnologías |
| Sumario: | This dissertation focuses on extensions of product line optimization modeling. Retail stores are faced with many decisions to optimize the process of satisfying customer demand. Thanks to current technologies, they can make use of diverse, relevant information in their decision making. Specifically, retail companies can incorporate into their optimization process information regarding the customer’s purchasing decisions. This type of problem structure can be modeled by bi-level programming, where the leader is represented by the retail firm and the followers are the customers. In particular, in this paper, we consider the product localization of a firm with multiple stores taking into account the customers’ reservation price and the cost of traveling to a store. In this way, customers may decide to travel further to purchase a cheaper product or buy a product that is not available in stores closer to them. By taking this into account, it makes the allocation of products more in line with consumer preferences, decreasing product turnover and avoiding higher inventory costs and/or product discounts. A second problem considered in this dissertation is a consumer coalition problem, where a set of consumers decide to group together to buy products taking advantage of wholesale prices. The products are available in baskets, and each customer has a different reservation price for them. We formulate both problems as bi-level optimization problems and explore cut generation strategies to solve these problems to optimality for real-world instances efficiently. In particular, for the product localization problem, we formulate it as an equivalent one-level problem and evaluate Lagrangian relaxation and cut generation methods to improve computational times. We introduce new cuts for this bi-level problem that improves computational times. Computational experiments suggest that valid inequalities reduce the linear relaxation gap, and embedded in Branch and Bound tree efficiently improve the best algorithm current -benders decomposition- known to the case of one store. For the customer coalition problem, the focus is to perform coalition formation. A general wholesale pricing function and step price function is presented. A Benders Decomposition is considered to solve large-scale instances. Computational experiments establish that adds the benders cut in root node has the best yield. |
|---|