Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

WD10

2 - Fairness Metrics in Facility Location Problems Gireeja Ranade, UC Berkeley, Atlanta, GA, Swati Gupta, Akhil Jalan, Helen Yang, Simon Zhuang We consider the NP-hard facility location problem: given a set of facilities and a set of consumers to serve, determine a subset of k-facilities that minimizes the sum of distances from each consumer to the nearest facility. For public facilities such as hospitals, playgrounds and waste collection facilities, it is important to take equity across groups into consideration. Various equity metrics have been proposed in the literature and our work explores the tradeoffs between them. 3 - Fairness in Prediction Models Used by Public Service Agencies Maria De-Arteaga, Carnegie Mellon University The use of machine learning to assist public service agencies is both promising and concerning, with problems ranging from child welfare to criminal justice and preventive health. While at first sight many of these may appear to be traditional prediction problems, the characteristics of available data and deployment contexts give rise to several challenges that have not been sufficiently addressed in the machine learning literature. Issues include presence of unobservables, selective labels, and omitted payoff biases. This talk discusses some of these challenges and presents novel methodologies to tackle them. 4 - Finding Equitable Start Times for Boston Public Schools Arthur J. Delarue, Massachusetts Institute of Technology, 77 Massachusetts Ave, Cambridge, MA, 02139, United States, Dimitris Bertsimas, Sebastien Martin Over 80% of high schools across the U.S. begin school before 8:30AM, even though early start times have been linked to serious health issues for teenagers. Because districts cannot estimate the impact of bell time changes on the cost of their complex bus schedules, change is often impossible under a fixed budget. We present a new optimization approach for the bell time assignment problem, which takes into account transportation costs as well as the complex needs of school districts. We discuss the application of our methodology as part of our collaboration with Boston Public Schools, which led to the unanimous approval of the first new start time policy in almost thirty years by the Boston School Committee. OM in Semiconductor Manufacturing II Sponsored: Manufacturing & Service Oper Mgmt Sponsored Session Chair: John W Fowler, Arizona State University, Tempe, AZ, 85287- 4706, United States Co-Chair: Lars Moench, University of Hagen, Hagen, 58097, Germany 1 - Design of Remote Diagnostic Networks for Semiconductor Equipment Suppliers John W. Fowler, Arizona State University, P.O. Box 874706, Supply Chain Managament, Tempe, AZ, 85287-4706, United States, Giulia Pedrielli, Esma S. Gel With advances in information technology, service activities for expensive equipment used in semiconductor manufacturing can be performed from a remote location. This is called remote diagnostics (RD). RD has the potential to reduce maintenance and capital costs and improve productivity. We develop a queueing-location model to analyze the capacity and location problem of after sales service providers, considering the effects of RD technology. Our model optimizes the location, capacity and the type of service centers while taking congestion effects into consideration. We solve this model using a simulation optimization approach in which we use a genetic algorithm to search the solution space. 2 - The Importance and Trends in Central Planning – Illusion of Control Kenneth J. Fordyce, Arkieva, 5460 Fairmont Drive, Wilmington, DE, 19808, United States Most organizations can be viewed as an ongoing sequence of loosely coupled decisions where current and future assets are matched with current and future demand across the demand-supply network at different levels of granularity. Since the middle 1990s, enterprise-wide supply, master or central planning, whose purpose is to create a projected supply linked to detailed exit demand and synchronize the activities of the enterprise, has become a key member of this “decision suite within the Kempf-Sullivan Decision Grid. This presentation provides an over of key trends such as increased granularity and diffusion and why it is important - even if it just creates the Illusion of control. n WD10 North Bldg 125A

3 - Improving Intel’s Supply Planning Performance Using a Proprietary Enterprise Optimization Solution Suite Shamin Shirodkar, Intel Corp, Chandler, AZ, United States, Zhenying Zhao As Intel has transitioned from High-Volume Low-Mix (HVLM) to High-Volume High-Mix (HVHM) manufacturing, supply planning has become much more complex & effective supply planning translates to a clear competitive advantage. In this presentation we are will be sharing an enterprise suite of proprietary well integrated optimization solutions which support supply planning at Intel & in the process various advanced generic modeling concepts, innovative formulations and solving techniques will be introduced.

n WD11 North Bldg 125B

Joint Session MSOM/Practice Curated: Smart City Operations Management Sponsored: Manufacturing & Service Oper Mgmt Sponsored Session Chair: Ho-Yin Mak, University of Oxford, Oxford, OX1 1HP, United Kingdom 1 - The Role of Consumer Waste in Food Retail Ekaterina Astashkina, INSEAD, Boulevard de Constance, Fontainebleau, 77305, France, Elena Belavina, Simone Marinesi We investigate how addition of consumer food waste changes the inventory dynamics and emissions of the upstream actors. We find that consumer food waste (unlike transportation emissions) generates spillover effects on upstream tiers: higher consumer food waste prompts higher retail sales and emissions. Hence, practices that cut consumer food waste are particularly efficient in contrast to other instruments that aim to benefit the environment. Further, we find that interventions targeted to reduce travel in the attempt to lower market emissions may in fact achieve the opposite result and increase market emissions, on account of higher consumer food waste. 2 - Revenue Management in Parking Systems under Competition and Information Asymmetries Xin Wang, University of Wisconsin, WI, United States, Qiao-Chu He, Yuguang Wu We consider oligopoly pricing game in parking systems under incomplete information. Each agent (garage) has some information (forecast) about the incoming parking demand for some of the garages. They charge a price for their own garage to maximize their profit. We consider the equilibrium wherein each agent’s action depends linearly on known forecasts and forms Bayes’ estimator for unknown forecasts. Based on the equilibrium strategy, we further investigate the agents’ incentive to share their information with others in order to get larger expected payoff. 3 - Peer-to-peer Crowdshipping as an Omnichannel Retail Strategy Ho-Yin Mak, University of Oxford, Park End Street, Oxford, OX1 1HP, United Kingdom We study a peer-to-peer crowdshipping strategy for an omnichannel retailer, who enlists in-store shoppers to deliver online orders in their vicinity. Using an analytical model, we investigate the effect of the crowdshipping model on the retailer’s profit and consumer surplus. We find that the favorability of crowdshipping heavily depends on the choice of reimbursement scheme and product characteristics. n WD12 North Bldg 126A Practice- Marketing I Contributed Session Chair: Frank Hage, Technical University of Munich, Production and Shengming Zheng, University of Science & Technology of China, 96 Jinzhai Road, School of Management, Hefei, 230026, China, Yugang Yu, Fuqiang Zhang, Xiaolong Guo Bundling strategy has been widely adopted by online retail platforms under agency selling. We study the optimal bundling strategy for a retail platform through which two independent manufacturers distribute their products. The manufacturers first set their product prices and then the platform decides whether to offer the bundled product or not and the price for the bundled product if it offers a bundle. By analyzing this two-stage Stackelberg game, we find that the platform will adopt the bundling strategy only when the manufacturer’s product prices are above a threshold. Additionally, the manufacturers will set high prices to induce the platform to offer bundled products. Supply Chain Management, Munich, 80333, Germany 1 - Optimal Bundling Strategy for a Retail Platform under Agency Selling

498