US-Turkey Cooperative Research: International: Bayesian Inventory Management under Censored Demand

项目描述：该奖项旨在支持一项合作研究项目，该项目涉及土耳其安卡拉比尔肯特大学的Richard Levine博士（加州戴维斯分校统计系）、Emre Berk博士（管理系）和Ulku Gurler博士（工业工程系）。这些科学家将研究贝叶斯更新方法的发展，以得出统计推断并推导出定期回顾库存问题的最优解。目标是开发现实的贝叶斯模型，用于库存管理应用中的决策，当1)库存需求不可观察或部分观察（销售损失）；2)顾客对即时库存和延迟库存的态度存在差异；3)需求过程随着时间的推移而演变。在库存管理的背景下，目标是通过利用先进的统计方法对需求过程中的不确定性进行建模，并结合贝叶斯方法在操作过程中随着数据的积累动态更新可用信息来实现。范围：该奖项将允许加州大学戴维斯分校的助理教授与两位土耳其科学家合作，其中一位也是最近的博士学位。该项目可能会增强长期富有成效的合作潜力。这个话题具有科学意义。其目的是弥合专注于给定系统优化的纯工业工程/运筹学方法与从给定数据集估计系统参数提供一次性解决方案的通用统计方法之间的差距。这一合作为处理这一复杂问题带来了所需的互补专业知识。该项目为美国和土耳其的这些领域带来了活跃的研究人员。该项目符合国际情报机构支持互利合作研究的标准。

0114738LevineDescription: This award is for support of a collaborative research project involving Dr. Richard Levine, Department of Statistics, University of California at Davis, Davis, California, Dr. Emre Berk, Department of Management, and Dr. Ulku Gurler, Department of Industrial Engineering, both at Bilkent University, Ankara, Turkey. These scientists will study the development of Bayesian updating methods to draw statistical inferences and derive optimal solutions in periodic review inventory problems. The goal is to develop realistic Bayesian models for decision-making in inventory management applications when1) inventory demands are unobservable or partially observed (lost sales); 2) customers differ in their attitudes toward immediate and delayed inventories; 3) demand processes evolve over time. In the context of inventory management, the goal is to be achieved by utilizing advanced statistical methods for modeling the uncertainty in demand processes and incorporating Bayesian methods to update available information dynamically during the course of operation as data accumulates. Scope: This award will allow an Assistant Professor at UC Davis to collaborate with two Turkish scientists, one of whom is also a recent Ph.D. This project is likely to enhance the potential for long-term productive collaboration. The topic is of scientific interest. The purpose is to bridge the gap between the purely industrial engineering/operations research approach that focuses on optimization of a given system and the common statistical approach that provides a one time solution for the estimation of system parameters from a given data set. The collaboration brings experts with needed complementary expertise to deal with this complex problem. The project brings active researchers in these areas in the United States and Turkey. The project meets the INT criteria for support of collaborative research of mutual benefit.