Informs Annual Meeting 2017

MB02

INFORMS Houston – 2017

3 - New Results on Risk Attitudes in Operations Decisions Andrea Hupman Cadenbach, University of Missouri-St. Louis, 211 Express Scripts Hall, 1 University Boulevard, St Louis, MO, 63121, United States, cadenbach@umsl.edu Previous results in the literature have shown that in repeated newsvendor decisions, the optimal ordering policy for a risk averse decision maker is approximately the same as for a risk neutral decision maker. These results, however, are based on an assumption of independent demand from one period to the next. In this talk, we present new results that show that in the presence of demand dependence, the optimal ordering policy for a risk averse decision maker differs from the optimal risk neutral policy. These results are extended to a general sequential decision making scenario. 4 - Theory of Generalized Risk Attitudes Janne Gustafsson, Ilmarinen Mutual Pension Insurance Company, Porkkalankatu 1, Helsinki, 00018, Finland, janne.gustafsson@gmail.com This paper develops a theory of risk attitudes that is based on the premise that a risk neutral decision maker (DM) is indifferent between a lottery and the average (in terms of preference) of the outcomes obtained from infinite repetition of the lottery. We show that a risk neutral DM seeks to maximize the expectation of measurable value. The results suggest that (i) the DM’s risk attitude in expected utility theory is related to the transformation between the measurable value function and the von Neumann-Morgenstern utility function and (ii) the applicability of the conventional definitions of risk attitudes is limited to settings in which the DM’s measurable value function is linear. 310B Beliefs in Environmental Decision-making Processes Sponsored: Decision Analysis Sponsored Session Chair: Melissa A Kenney, University of Maryland, College Park, MD, 20740, United States, kenney@umd.edu 1 - Finding Common Ground when Experts Disagree: Energy and the Environment Erin Baker, University of Massachusetts-Amherst, MIE, 220 Elab, Amherst, MA, 01003, United States, edbaker@ecs.umass.edu, Ahti Salo, Valentina Bosetti We address the problem of decision making under “deep uncertainty,” introducing an approach we call Robust Portfolio Decision Analysis. We derive a set of portfolios that are non-dominated across beliefs; and then identify robust individual alternatives from the non-dominated portfolios. The process allows us to synthesize multiple expert- or model- based beliefs by uncovering the range of alternatives that are intelligent responses to the range of beliefs. We show that this method encompasses many of the robustness concepts in the literature. We illustrate our approach using a problem in the climate change and energy policy context: choosing among clean energy technology R&D portfolios. 2 - Effect of Decision Rules on Chemical Alternatives Assessment Michael Gerst, University of Maryland, 1337 Meridian Place NW, Washington, DC, 20010, United States, mgerst@umd.edu, Melissa A.Kenney, Brett Howard Alternatives assessment (AA) is an overarching term for the process of identifying, comparing, and selecting safer alternatives to chemicals that might pose a risk to humans or the environment. To date, many methods have been produced, each with foci that largely align with the discipline producing the method. We demonstrate the efficacy of a multi-criteria approach in an AA of 14 paint stripper chemicals by showing its flexibility compared to other decision rules. 3 - Characterizing the Evolution of Global Ground-water Stress Roshanak Nateghi, Characterizing Global Ground Water Storage Trends, Purdue University, West Lafayette, IN, 47906, United States, rnateghi@purdue.edu, C. Bayan Bruss, Benjamin Zaitchik Groundwater is a key element of global access to water resources; and has important implications for political stability, economic growth and public health in every society. Recent rapid population growth, and increased rates of urbanization have led to groundwater stress. Identifying the national-scale drivers of groundwater storage trends is the key first step for understanding water availability and devising sustainable water management policy in a changing climate. In this research we have conducted extensive data analyses to find the key predictors of the observed groundwater storage trends using agricultural, climate, demographic, land-use and economic variables in 81 countries. MB02

4 - Decision Analysis Approach to Environmental Model Interpretation

Melissa A. Kenney, Associate Research Professor, University of Maryland, 5825 University Research Court, Suite 4001, College Park, MD, 20740, United States, kenney@umd.edu, Eva D. Regnier, Michael Gerst Bayesian reasoning - asking what model output we ought to expect as a function of the possible realities - is rarely applied to the interpretation of the output from mechanistic models. Specifically, our framework treats the model as an evidence- generating process from which the analyst considers the joint relationship between reality and model output. This process yields a likelihood function that (i) serves to guide what questions should be asked in the model building and validation process and (ii) how insights should be updated given prior information and model outputs. To illustrate the approach, we discuss physical climate models and integrated assessment models.

MB03

310C Humanitarian Logistics

Invited: Tutorial Invited Session Chair: Jiming Peng, University of Houston, Houston, TX, 77204, United States, jopeng@Central.uh.edu Co-Chair: Rajan Batta, University at Buffalo (SUNY), 410 Bell Hall, Buffalo, NY, 14260, United States, batta@buffalo.edu 1 - Humanitarian Logistics Bahar Yetis Kara, Bilkent University, Ankara, Turkey, bkara@bilkent.edu.tr, Sinem Sava er Many incidents requiring humanitarian assistance, which can be categorized mainly as disasters and long term crises, have taken place in the recent years. Humanitarian logistics play a crucial role in managing these events because of the response time constraints, limited resource amounts and high levels of uncertainty. Therefore, it is apparent that the challenges faced in humanitarian logistics differ from the conventional challenges encountered in commercial logistics applications such as having conflicting goals for different parties, funding and political issues. In this tutorial, we provide an introduction to humanitarian logistics by defining its characteristics, application areas and challenges. Then, we analyze and synthesize relief logistics and development logistics literature based on Operations Research (OR) problems encountered in disaster management cycle while putting emphasis on the last 10 years. For various disaster management stages; namely preparedness and response phases, certain case studies will be discussed using a real life dataset from Kartal region of Istanbul. Moreover, for development logistics, a case study on providing healthcare in rural areas will be evaluated. MB03A Grand Ballroom A Cloud Computing and Data Centers in Applied Probability I Sponsored: Applied Probability Sponsored Session Chair: Alexander Stolyar, University of Illinois at Urbana-Champaign, Urbana, IL, 61801, United States, stolyar@illinois.edu 1 - On TTL Caching in Content Delivery Networks Sanjay Shakkottai, University of Texas-Austin, ENS.437 ECE Department, Austin, TX, 78712, United States, shakkott@austin.utexas.edu, Soumya Basu, Aditya Sundarrajan, Javad Ghaderi, Ramesh Sitaraman We study TTL caching in content delivery networks (CDNs). Using a stochastic approximation approach, we develop algorithms that adapt the time-to-live parameter to current workloads, and demonstrate performance gains over the state-of-art. 2 - Control of Parallel Server Systems with Learning Yuan Zhong, University of Chicago, Booth School of Business, Chicago, IL, 60637, United States, Yuan.Zhong@chicagobooth.edu, Jean Walrand We consider parallel server processing systems, where the service rates have dependence on both the job class and the server, and we are interested in control policies that do not assume knowledge of system parameters. First, we show that the classical max-weight policy, modified under certainty equivalence, is not throughput optimal, hence the necessity of forced learning. Then, we propose two classes of control policies that incorporate learning in different ways and prove that both classes are throughput optimal.

164