Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

MC75

4 - A New Online Tool to Comprehensively Support Multi-Criteria Decision Aiding Processes: ElectioVis Maximiliano Ariel López, University of Buenos Aires, Buenos Aires, Argentina, Valentina Ferretti This research proposes the development of a new online Multi-Criteria Decision Aiding tool that is able to support expert and non-expert users. It assists the method selection phase (i.e. choosing the best tool given the characteristics of the decision opportunity under analysis) as well as the model deployment (identifying the best solution for that specific decision). The interactive support provided by the ElectioVis online decision support system bridges insights from behavioural and normative theories to allow for the development of more successful interventions and solutions. n MC75 West Bldg 212B Joint Session MAS/Practice Curated: Analytic and Data Science Applications Sponsored: Military and Security Sponsored Session Chair: Jon Alt, Naval Postgraduate School, Monterey, CA Co-Chair: Eric Tollefson 1 - Leveraging the Cloud Computing Environment to Support Decision Makers Nathan I. Parker, TRADOC Analysis Center - Monterey, 700 Dyer Road, Room 178, Monterey, CA, 93943, United States, Jonathan Shockley Here we use our ongoing effort to develop a cloud-hosted, browser interfaced decision support tool (DST), leveraging one of TRADOC Analysis Center’s simulation models, as a starting point to discuss how cloud computing provides a unique capability to enable distributed data science applications. In addition to presenting the Logistics Battle Command DST itself, we will also discuss the underlying cloud infrastructure we employ and our development, test, and deployment pipeline. 2 - Aerial Exposure Metric James Jablonski, TRADOC Analysis Center, 1106 Leahy Rd, Monterey, CA, United States The development of a method to calculate the level of exposure of each point in the sky in a given terrain box to points on the ground. The exposure metric can then be used to characterize the exposure of an area, to determine the optimal routing to minimize exposure in a given area, or to determine the optimal placement of air defense assets to maximize coverage in an area. The research also informs the computational costs of scaling these methods up to a larger terrain box and provides insights into future directions. A proof of principle application, developed using open source tools is provided. 3 - Binary Classification with Asymmetric Error Type Control Mathew Norton We introduce a new formulation for binary classification with asymmetric error control inspired by the Neyman-Pearson (NP) paradigm. We propose a computationally efficient large margin classifier with the same generalization benefits as Support Vector Machines in high-dimensional feature spaces, but with fine tuned control over the amount of allowable Type I and Type II error. Our approach is based on a new characterization of uncertainty called Buffered Probability of Exceedance (bPOE) and, as consequence, often reduces to convex or linear programming. 4 - A Methodological Framework for Developing a Defense Data Strategy Kurt Klingensmith, TRAC, Monterey, CA, United States, Jon Alt In order to lead the Army’s modernization, the emerging Army Futures Command (AFC) will leverage timely, relevant, and credible analysis to informs and drive critical modernization strategies, priorities, and decisions. Achieving this requires the development and operationalization of a formal AFC Data Strategy. The Data Strategy envisions how the AFC will use data as an enterprise asset in support of modernization activities such as concept development, modernization planning, acquisition, and capability delivery. This presentation will present a systems architecting-based methodology for developing data strategies along with a framework for implementing and executing a data strategy.

n MC73 West Bldg 211B JFIG Panel Discussion: Tips for Successful Publication from Journal Editors Sponsored: Junior Faculty JFIG Sponsored Session Chair: Anahita Khojandi, University of Tennessee, Knoxville, TN, 37996, United States 1 - JFIG Panel Discussion: Tips for Successful Publication from Journal Editors Anahita Khojandi, University of Tennessee, 521 Tickle Building, Knoxville, TN, 37996, United States Past and current editors from top journals will share tips on how to get your paper successfully published, from selecting the right journal to preparing and finalizing the manuscript. Panelists Joseph Geunes, Texas A&M University, College Station, TX, United States Jianjun Shi, Georgia Institute of Technology, H. Milton Stewart School of, Industrial and Systems Eng, Atlanta, GA, 30332-0205, United States Baris Ata, University of Chicago, Booth School of Busines, 5807 S. Woodlawn Ave, Chicago, IL, 60637, United States Eva Lee, Georgia Tech, Industrial & Systems Engineering, Ctr for Operations Research in Medicine, Atlanta, GA, 30332-0205, United States Preference-driven Decision Aiding I Sponsored: Multiple Criteria Decision Making Sponsored Session Chair: Roman Slowinski, Poznan University of Technology, 60-965, Polandl Co-Chair: Adiel Teixeira De Almeida Filho, Universidade Federal de Pernambuco, Recife-PE, 50.630-971, Brazil 1 - On Rationality Conditions for Multi-attribute Choice Behavior Pekka J. Korhonen, Aalto University School of Business, Pahkinaranta 10, Huhmari, 03150, Finland, Jyrki Wallenius, Peng Xu, Tolga Genc This paper deals with rationality conditions for choice behavior. We explore two different types of choice settings: (1) Win-Win setting, (2) Tradeoff setting. We study the decision-maker’s rationality conditions in both settings. The key underlying theoretical assumptions in our paper are increasing and concave single-dimensional utility (value) functions with decreasing marginal values and the Kahneman-Tversky Prospect Theory model of choice with loss aversion. We use an empirical experiment to illustrate our considerations. 2 - Shared Decision Making (SDM): A Preference-driven Approach to Medical Decision Making Evangelos Triantaphyllou, Louisiana State University, Dept of Computer Science, 298 Coates Hall, Baton Rouge, LA, 70803, United States, Edouard Kujawski, Juri Yanase A unique and important type of decision making occurs when a patient and clinicians collaborate to decide the best treatment. SDM is a fast emerging preference-driven approach. It aims at determining the treatment that best meets the personal values, goals and preferences of the patient. However, some SDM approaches may be seriously defective as they rely on erroneous models. 3 - Evolutionary Multiobjective Optimization Guided by Preferences Modeled with Achievement Scalarizing Function Roman Slowinski, Professor, Poznan University of Technology, Pl. Marii Sklodowskiej-Curie 5, NIP: 777-00-03-699, Poznan, 60-965, Poland, Tomasz Sternal A new preference-driven interactive method is proposed that organizes the evolutionary search of multiobjective space along a set of directions indicated by a preference model consistent with some preference information elicited from the Decision Maker (DM). It is compared in a computational experiment to state-of- the-art multiobjective evolutionary optimization algorithms MOEA/D and NSGA-III. Our method applies an augmented Chebyshev function as a preference model. Adjusting the model parameters relies solely on a number of pairwise comparisons of some non-dominated solutions elicited from the DM during the algorithm run. n MC74 West Bldg 212A

211