A General Framework for Incorporating Choice Models into Operations Management Decisions

许多公司倾向于为客户提供种类繁多的产品，以增加客户基础。例如，人们可能在零售商那里看到二十种或更多型号的数字设备，每种型号都在功能和价格上相互竞争。同样，在线旅游网站也提供各种各样的旅游产品，这些产品来自不同的供应商。选择合适的产品种类来提供给客户是一个难题。提供低利润率的产品可以增加客户基础，但也可能导致现有客户转向低利润率的产品，从而损害收入。大的产品种类增加了客户群，但也可能增加运营成本。随着库存的消耗或公司对客户选择的了解越来越多，可能有必要调整提供的分类。尽管选择无处不在，但运营决策和客户选择之间的接口并没有得到很好的理解。本研究的主要目标是开发一个将客户选择集成到运营管理决策中的框架。本研究将集中于客户选择行为的静态和动态优化问题。在静态设置中，底层选择模型的参数是已知的，目标是决定提供哪种产品分类。在动态环境下，底层选择模型的参数是不确定的。人们必须根据数据估计参数值，并随着参数值知识的发展而调整决策。除了方法上的贡献外，本研究将影响库存控制、供应链管理和收益管理的相关课程设置。将开发模拟游戏来验证研究结果，并将其用作教学平台。来自不同背景的各级学生将有机会获得研究经验。

Many companies favor offering a large product variety to their customers with the intention of increasing the customer base. For example, one may see twenty or more models of a digital equipment at a retailer, each model competing with others in features and price. Similarly, online travel websites offer a variety of travel products consolidated from different suppliers. Choosing the right variety of products to offer to the customers turns out to be a difficult problem. Offering a low-margin product increases the customer base, but it may also cause the current customers to switch to the low-margin product, damaging revenues. Large product variety increases the customer base, but it may also increase the operational costs. As the inventories are consumed or as the company learns more about the customer choices, it may be necessary to adjust the offered assortment. Despite the ubiquitous nature of choice, the interface between the operational decisions and the customer choices are not well-understood. The primary goal of this research is to develop a framework for integrating customer choice into operations management decisions.This research will focus on both static and dynamic optimization problems with customer choice behavior. In the static setting, the parameters of the underlying choice model are known and the objective is to decide which assortment of products to offer. In the dynamic setting, the parameters of the underlying choice model are uncertain. One must estimate the parameter values based on data and adapt the decisions as the knowledge of the parameter values evolves. Besides methodological contributions, this research will impact the relevant course offerings in inventory control, supply chain management and revenue management. Simulation games to validate the findings will be developed and used as an instruction platform. Students at all levels and from diverse backgrounds will have opportunities for research experience.