Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

TE81

4 - Product Line Planning Under Uncertainty Sakir Karakaya, Industry and Technology Expert, Ministry of Science, Industry and Technology, Mustafa Kemal Mah. Eskisehir Yolu 7.km, Ankara, 06510, Turkey, Gulser Koksal This study addresses the problem of multi-period mix of product-lines under a product family, which incorporates launching decisions of new products, capacity planning decisions and product interdependencies. The problem is modelled as a two-stage stochastic program in which price, demand, production cost and cannibalisation effect of new products are characterised as uncertain parameters. The solution approach employs the Monte Carlo bounding technique and L- shaped method, which is tested on real-like cases considering VSS and EVPI performance measures. The data collected through two experimental studies is analysed using ANOVA and Random Forest methodology. 5 - A Comparison of Stochastic Service Model and Guaranteed The Stochastic Service Model (SSM) and Guaranteed Service Model (GSM) are competing models for optimizing multi-echelon inventory systems. SSM accounts for the uncertainty in lead times due to stockouts, while GSM treats lead times as bounded. SSM is more representative of real-world supply chains, but GSM is more tractable and widely used in practice. Our research asks whether GSM is an effective heuristic for SSM. Our results suggest that SSM and GSM agree closely in serial and assembly systems under many conditions. However, the two models often diverge for distribution systems, making GSM more problematic for more complex topologies. 6 - How Does Conspicuous Consumption Affect Product Line Design Mengke Tian, the Hong Kong University of Science and Technology, Hong Kong, Ying-Ju Chen, Xin Wang We analyze the effects of conspicuous consumption on the product line design when there is only a monopoly manufacture who tends to directly serve the market with high and low-end products. We find that stronger conspicuous effect undercuts the quality and prices for both products which leads to the increase of demands. The total profit of the manufacture is elevated. The gaps of quality standard and selling price are widened. We extend by adding a retailer in the distribution channel. We find that manufactures adopts the same quality strategies as before. The double marginalization effect leads to higher prices and lower demands. But its influence is mitigated when conspicuous consumption is stronger. n TE81 Hyatt, Phoenix East Robust Optimization Contributed Session Chair: Duaa Serhan, Binghamton University, Binghampton, NY, United States 1 - From Data to Multi-stage Decisions In this talk, we consider multi-stage linear optimization problems, where the uncertainty is correlated between stages and our only information comes from historical data. We propose a simple data-driven framework, which we refer to as “sample robust optimization” (SRO). First, we establish that, as more historical data is obtained, SRO prescribes decision rules with (asymptotically) optimal out- of-sample performance. Second, exploiting structural similarities between SRO and robust optimization, we present practically-tractable approaches for solving SRO problems. Finally, we discuss empirical experiments and compare to approaches from distributionally robust optimization. 2 - Optimization Over Continuous and Multi-dimensional Decisions with Observational Data Christopher G. McCord, Massachusetts Institute of Technology, 77 Massachusetts Ave, Cambridge, MA, United States, Dimitris Bertsimas We consider optimization under uncertainty over continuous and multi- dimensional decision spaces in problems in which we are only provided with observational data. We propose a novel algorithmic framework that is tractable, asymptotically consistent, and superior to comparable methods on example problems. Our approach leverages highly effective predictive machine learning methods for the purpose of prescribing decisions. We demonstrate the efficacy of our method on examples involving both synthetic and real data sets. 3 - A Nested Robustness Approach for the Mobile Vendor Location Problem Juan Carlos Espinoza Garcia, Assistant Professor, Tecnologico de Monterrey, Queretaro, 76130, Mexico, Laurent Alfandari We propose an operational-level variant of the Capacitated Facility Location Problem applicable to mobile vendors, where the demand to each potential Service Model in Multi-echelon Inventory Systems Yinan Liu, Lehigh University, Philadelphia, PA, 19144, United States Bradley E. Sturt, Massachusetts Institute of Technology, Cambridge, MA, 02139, United States, Dimitris Bertsimas, Shimrit Shtern

location is a function of the selected locations. For this problem (MVLP) the demands follow a Poisson process, which provides fractional functions in the objective of the robust model when considering uncertainty on demands. We introduce the concept of Nested Uncertainty Budgets to address overly conservative solutions and propose a method to manage global and local uncertainty in a hierarchical way. Numerical experiments are provided on adapted benchmark instances to assess the performance of the nested robust model and a classical approach. 4 - Robust Optimization for Routing and Scheduling of Airport Surfaceand Terminal Airspace Operations under Pushback and Arrival Times Uncertainty Duaa Serhan, Binghamton University, Binghamton, NY, 13902, United States, Sang Won Yoon This research proposed robust optimization approach for scheduling and routing airport surface and terminal airspace (ASTA) operations under uncertain fight schedules. Convective weather is the leading cause of flight delay in the National Airspace System. This paper aims to develop a decision support tool for routing and scheduling of ASTA operations during convective weather and different levels of aircraft pushback and arrival time uncertainty. A robust optimization approach is proposed to identify aircraft routes and schedules that minimize flight delays and safely avoid areas of convective weather. The proposed approach is expected to safely and efficiently manage aircraft operations. 5 - An Economic Model Explaining Asymmetric Price Responses Dong Soo Kim, Assistant Professor of Marketing, Ohio State University, 500 Fisher Hall, 2100 Neil Ave., Columbus, OH, 43210, United States, Mingyu Joo, Greg Allenby This paper proposes a model that considers both expectation and uncertainty of information that formulates reference points as a distribution of outside option prices in a utility maximization framework. The model assumes that consumers compare the expected maximum attainable utility of outside options with the marginal utility from inside goods with observed prices. In doing so, the model characterizes the opportunity cost as a certainty-equivalent value, a value- adjusted standard to compare against. It shows that reference prices operationalized as point estimates in the literature may over-estimate the actual subjective standard, leading to asymmetric gain and loss parameter estimates. n TE82 Hyatt, Phoenix West Health Care V Contributed Session Chair: Stephen J. Stoyan, Abbott Laboratories, 100 Abbort Park Road, Chicago, IL, 60064, United States 1 - A Dynamic Decision Framework of Vaccine Distribution for Influenza Interventions Lo-Min Su, National Chiao Tung University, Hsinchu, Taiwan, Yu-Hsuan Wu, Sheng-I Chen This study emphasizes on the vaccine distribution decision for national immunization programs with the consideration of uncertain epidemic pattern. We integrate disease and supply chain models, where a compartmental model is developed to provide the estimations of transmission intensity and population settlements, and a mixed-integer programming model to determine vaccine allocation according to the epidemic result at each stage. A forecast model of vaccine demand is also developed to acquire the demand for the integration model. We investigate the effectiveness of centralized and decentralized vaccine inventory policies under irregular pandemic and regular epidemic. 2 - Grounding Frequent Flyers: Examining Utilization under the Patient Protection and Affordable Care Act Eric Xu, University of Minnesota, Minneapolis, MN, United States With the passage of the Patient Protection and Affordable Care Act, policy makers assumed that insurance coverage would encourage individuals to use primary care instead of emergency department care. One of the key tools to achieve this goal was Medicaid Expansion, which provided coverage to most vulnerable populations. However, under Medicaid Expansion, it appears that the utilization of ED and primary care services is rising. This research seeks to understand how the policy has impacted care delivery and utilization, in addition to capacity planning initiatives moving forward. 3 - The Effect of Lifestyle Choices on Cancer: Which One has the Strongest Influence? Banafsheh Behzad, California State University, Long Beach, CA, United States Cancer is known to be caused by both genetic and environmental factors, with 80% of cancers being related to environmental factors. This shows the significance of our lifestyle choices in preventing cancer. The main lifestyle choices that affect the occurrence and mortality of cancer, analyzed in this study, are alcohol consumption, lack of physical activity, obesity, and tobacco usage. The strength of the association between cancer and these lifestyle choices are studied. The results show that the most influential lifestyle choices in predicting cancer are alcohol consumption and obesity.

408