Data-Driven Revenue Management : The Interplay of Data, Model, and Decisions

Revenue management (RM) is the application of analytical methodologies and tools that predict consumer behavior and optimize products' availability and prices to maximize a firm's revenue or profit. In the last decade, data has been playing an increasingly crucial role in business decision-making. As firms rely more on collected or acquired data to make business decisions, it brings opportunities and challenges to the RM research community. In this review paper, we systematically categorize the related literature by how a study is "driven" by data and focus on studies that explore the interplay between two or three of the elements: data, model, and decision, in which the data element must be present. Specifically, we cover five data-driven RM research areas, including inference (data to model), predict-then-optimize (data to model to decision), online learning (data to model to decision to new data in a loop), end-to-end decision making (data directly to decision), and experimental design (decision to data to model). Finally, we point out future research directions