管理科学与工程系学术讲座系列2021年第2讲

主 题：Offline-Channel Planning in Omnichannel Retail

主讲人: 沈浩 (中国人民大学商学院讲师)

时 间：2021-04-14 10:30

地 点：明商706会议室

语 言：中英文

讲座摘要：

Problem definition: Observing the retail industry inevitably evolving into omnichannel, we study an offline-channel planning problem that helps an omnichannel retailer make offline-channel planning decisions, given that customers’ purchase decisions depend on not only their preferences across products but also their valuation discrepancies across channels, as well as the hassle costs incurred. In particular, in the presence of an online store, we address how a retailer should situate offline stores and determine the offline location-dependent assortment in each store to maximize revenue across both online and offline channels.

Academic / Practical Relevance: The proposed model builds on the mixed multinomial logit choice model to optimize location-dependent offline assortments under the omnichannel setting while incorporating other important planning decisions such as the location of offline stores to cater to omnichannel retailers. The model and the solution approach also extend the literature on retail channel management, omnichannel assortment planning, and the broader field of smart retailing/cities.

Methodology: We derive parameterized models to capture customers’ channel choice and product choice behaviors, and demonstrate a corresponding parameter estimation approach. Based on this choice model, we propose an optimization model to guide offline store location and assortment planning decisions. Since the proposed model is NP-hard, we develop a tractable mixed-integer second-order conic program (MISOCP) reformulation and explore the structural properties of the reformulation to derive strengthening cuts. We show that these cuts can be generated in closed-form, which further improves solution efficiency.

Results: We analyze special cases of the location-assortment optimization model to obtain structural properties of the optimization problem and the corresponding optimal solutions. We also conduct numerical experiments to validate the necessity and efficacy of the proposed solution approach. Furthermore, based on real data sets, we demonstrate the parameter estimation approach, and apply the proposed solution approach to solve the optimization model and draw insightful observations.

Managerial Implications: We find that incorporating spatial heterogeneity of customer preferences improves revenues for retailers. Moreover, we reveal the main reasons why omnichannel retailers should provide location-dependent offline assortment. We also discover that offering products according to revenue order can be suboptimal.

主讲人简介：

Hao Shen is an assistant professor of Operations Management at School of Business, Renmin University of China. He received his Ph.D. in Management Science and Engineering, and a B.E. in Engineering Mechanics, both from Tsinghua University. His research interests include supply chain management and data-driven decision methods. His work has been published in Manufacturing & Service Operations Management, Production & Operations Management, and INFORMS Journal on Applied Analytics. He also has industry experience in JD.com and DiDi (as a Research/Data Scientist intern). A previous version of the work to be presented in this seminar has been recognized as the Winner of the POMS-JD.com Best Data-Driven Research Paper Competition.