Inventory control, revenue management and pricing: Data-driven stochastic optimization approaches

背景供应链管理在加拿大和全球经济中发挥着重要作用，随着系统的复杂性和产品的扩散，供应与需求的匹配变得越来越困难。随着信息系统的兴起，企业可以更好地了解供应网络和客户决策。虽然这种新兴数据有可能彻底改变供应链管理，但由于其规模，类型和速度，它也带来了新的挑战，要求企业采用新的方式来获得洞察力并控制供需过程。 目标和目的。我们的研究目标是通过业务分析和数据科学为供应链的整体绩效做出贡献。我们将提出和分析模型，捕捉供应系统中的基本权衡，并开发数据驱动的方法。重点是将数据更直接地纳入决策过程。 更具体地说，（a）我们将为库存系统开发数据驱动的方法，处理诸如审查需求信息、多供应层、易腐性和风险状况等问题。(b)我们还将研究固定产品和快速过渡产品的客户行为，将新兴的选择模型纳入公司的定价和分类决策，在学习和盈利（探索和利用）之间取得平衡。此外，（c）我们将研究如何将数据驱动的供应链管理原则应用于加拿大的医疗保健和运输行业。 科学方法。我们将使用三个方面的分析。 (a)描述性分析：对行业合作伙伴的观察和访谈;描述性统计;统计推断。 (b)预测分析：回归;模拟;分类树;聚类;机器学习。 (c)规范分析：随机最佳化;动态规划;启发式方法。 预期结果。我们将通过出版物、学术会议和面向制造业的行业委员会传播我们的研究结果。博士生将被培训为研究供应链的合格学者，其他学生将在私营部门或政府中支持数据分析。(HQPs占预算的大部分）。 对加拿大和市场有利。数据驱动的供应链管理正成为运营能力和竞争力的组成部分。通过更好地利用数据，产品和服务提供商将更好地管理流程，从而提高盈利能力和竞争力。在商业分析方面不擅长的公司将落后。最终客户将享受更好的产品和服务体验。政府决策者将深入了解监管替代方案的影响，因为加拿大国内生产总值的十分之一与供应链活动有关。

Background. Supply chain management plays an important role in the Canadian and global economy, and matching supply with demand is becoming increasingly more difficult with system complexity and product proliferation. With the rise of information systems, firms can have more visibility into supply networks as well as customer decision-making. While this emerging data holds the potential to revolutionize supply chain management, it also poses new challenges due to its size, type and velocity, requiring firms to adopt new ways to obtain insight and to control supply-demand processes. Goal and Objective. Our research goal is to contribute to the overall performance of supply chains through business analytics and data science. We will propose and analyze models that capture essential trade-offs in supply systems, and develop data-driven methods. The emphasis is given to incorporating data more directly into the decision making process. More specifically, (a) we will develop data-driven methods for inventory systems, handling issues such as censored demand information, multiple supply layers, perishability, and risk profiles. (b) We will also study customer behaviour, both for stationary products and fast-transitioning products, to incorporate emerging choice models into the firm's pricing and assortment decisions, striking a balance between learning and earning (exploration and exploitation). Furthermore, (c) we will examine how the principles of data-driven supply chain management can be applied to Canada's health care and transportation industries. Scientific Approach. We will use three aspects of analytics. (a) Descriptive analytics: Observation and interviews with industry partners; descriptive statistics; statistical inference. (b) Predictive analytics: regression; simulation; classification trees; clustering; machine learning. (c) Prescriptive analytics: stochastic optimization; dynamic programming; heuristic methods. Anticipated Outcome. We will disseminate our findings through publications, academic conferences and practitioner-oriented industry councils. Doctoral students will be trained as competent scholars in researching supply chains, and other students will champion data analytics in the private sector or the government. (HQPs account for a majority of the budget.) Benefit to Canada and the Marketplace. Data-driven supply chain management is becoming integral to operational competency and competitiveness. Through better use of data, product and service providers will better manage processes, resulting in higher profitability and competitiveness. Firms that do not excel in business analytics will lag behind. End-customers will enjoy better product offerings and service experiences. Government policy makers will gain insight into the impact of regulatory alternatives as one-tenth of Canada's gross domestic product is tied to supply chain activities.