題 目：在線季節性銷售的最優策略
主講人： 劉方 助理教授 (南洋理工大學商學院)
時 間：2017年12月19日 上午10:00
地 點：主樓317
主講人介紹：
Liu Fang is an assistant professor in the department of Information Technology and Operations Management, Nanyang Business School. She received her doctoral degree in Operations Management from the Fuqua School of Business, Duke University and her Bachelor’s degree from the School of Mathematics, Peking University. She has presented her research in conferences such as INFORMS, MSOM and POMS, and has published papers in Operations Research, Manufacturing & Service Operations Management and Production and Operations Management.
內容介紹：
Online retailing has been continuously growing. One of the biggest challenges for online retailers to run seasonal sales is to operate a responsive supply network such that they are able to efficiently fulfill extremely large demands from different zones (such as countries or regions) during a short selling season. We consider an on-line retailer selling multiple products and study the ordering, storage and retrieval decisions for seasonal sales. The on-line retailer first builds up her inventory by ordering the products from her supplier and then stores them to multiple warehouses at different locations. Finally, these products are used to fulfill the demands from different zones. For the single demand zone problem, the optimal retrieval policy can be characterized in closed form by solving independent knapsack problems. The optimal storage policy can be obtained using a general greedy algorithm that allocates the products to the warehouses iteratively according to each warehouse's updated target stock-out probability. Finally, the optimal ordering policy is a newsvendor-type policy. For the multiple-zone problem, we propose two heuristics: virtual demand pooling heuristic and virtual capacity heuristic. Virtual demand pooling heuristic controls the over-stocking risk and results in low purchase and storage costs. Virtual capacity allocation heuristic stores the products to the warehouses closer to their target zones and yields a low retrieval cost. A case study based on data from a major online retailer in Asia supports the above observations and reveals that a hybrid heuristic combining the strengths of the two heuristics outperforms all the other heuristics, and attains 91% efficiency compared to the expected value given perfect information.
（承辦：管理科學與物流系，科研與學術交流中心）
