Unified Framework for Choice-Based Facility Location Problem
针对基于选择的设施选址问题中模型众多、选择困难，本文提出基于偏好支配的统一建模框架，将顾客选择视为两步过程，并设计高效精确算法，在多种问题上优于专用方法。
The choice-based facility location (CBFL) problem arises in various industrial and business contexts. The problem stands on a decentralized perspective: Companies set up chains of facilities, and customers determine from which chain or facility to seek service according to their own preferences. Essentially, customer preferences or choices play a key role in characterizing various CBFL problems, which differ mainly in the models or rules used to characterize the choice. Consequently, a large number of formulations appear and are often solved by dedicatedly designed approaches in the literature. Such a situation significantly complicates practitioners’ decision-making process when they are facing practical problems but are unsure which ad hoc model is suitable for their cases. In this article, we address this dilemma by providing a unified modeling framework based on the concept of preference dominance. Specifically, we conceptualize the choice behavior as a sequential two-step procedure: Given a set of open facilities, each customer first forms a nondominated consideration set and then splits the buying power within the set. Such an interpretation renders practitioners high modeling flexibility as they can tailor how preference dominance is constructed according to their specific contexts. In particular, we show that our model can represent several streams of CBFL problems. To support our model’s applicability, we design an efficient exact decomposition algorithm. Extensive computational studies reveal that although the algorithm is designed for a general purpose, it outperforms most approaches that are tailored for ad hoc problems by a large margin, which justifies both the effectiveness and the efficiency of the unified framework. History: Accepted by Pascal Van Hentenryck, Area Editor for Computational Modeling: Methods & Analysis. Supplemental Material: The software that supports the findings of this study is available within the paper and its Supplemental Information ( https://pubsonline.informs.org/doi/suppl/10.1287/ijoc.2022.0366 ) as well as from the IJOC GitHub software repository ( https://github.com/INFORMSJoC/2022.0366 ). The complete IJOC Software and Data Repository is available at https://informsjoc.github.io/ .