Informs Annual Meeting 2017

TC33

INFORMS Houston – 2017

TC33

2 - Second-order Extensions to Nearly Orthogonal-and-balanced Mixed-factor Experimental Designs Zachary C.Little, Air Force Institute of Technology, Wright-Patterson AFB, OH, United States, zach.c.little@gmail.com Zachary C.Little, The Perduco Group, Beavercreek, OH, United States, zach.c.little@gmail.com, Jeffery D. Weir, Raymond R. Hill, Brian B.Stone, Jason K.Freels Many simulation studies can involve a large number of both quantitative (how many?) and qualitative (which type?) decisions with different numbers of choices for each. Second-order extensions are developed for the nearly orthogonal-and- balanced (NOAB) design that accounts for such varied decisions, allowing analysts to better examine possible two-way interactions and quadratic effects. A case study is presented for evaluation and comparison of NOAB design construction approaches for a design space involving notional Intelligence, Surveillance, and Reconnaissance (ISR) decisions. 3 - Data Science Application for Medical Intelligence” Russell Walter, U.S. Army, Washington, DC, 22207, United States, walt@mail.mil We examined machine learning and natural language processing to detect articles containing relevant warnings of disease outbreak. For years, medical intelligence analysts have categorized articles based on relevant information they contained, yielding an extensively labelled dataset to train a machine learning classifier. Differences between classes are subtle. Applying several machine learning algorithms, the biggest sensitivity resulted from the preparation of the training corpus and not the models’ parameters. Lessons learned regarding data having large class imbalances are also shared. 351E Supply Chain Management, Green Contributed Session Chair: Elif Elcin Gunay, Iowa State University, Ames, IA, United States, eekabeloglu@gmail.com 1 - Optimality of Rule-based Allocation Planning in Customer Hierarchies with Stochastic Demand wnd Heterogeneous Service-level Targets Konstantin Kloos, Research Assistant, University of Wuerzburg, Stephanstraße 1, Wuerzburg, 97070, Germany, konstantin.kloos@uni-wuerzburg.de, Benedikt Schulte, Richard Pibernik To account for the multi-level hierarchical structure of sales organizations companies often times refrain from centralized optimization-based approaches but use simple allocation rules (i.e. per commit, rank based,...) better suited for decentralized, top-down planning regimes. Using the global optimum as a benchmark, we show analytically that these simple rules are optimal only under very restrictive conditions. Based on our analytical insights we develop two new rules resulting in optimal allocations under less restrictive conditions. In a numerical experiment we show that these new approaches consistently out perform conventional simple rules and lead to close-to-optimal results. 2 - Optimal Product Architecture and Supply Chain Configuration Designs under Labor Cost Fluctuation Elif Elcin Gunay, Iowa State University, Ames, IA, United States, eekabeloglu@gmail.com, Kijung Park, Gul Kremer Many companies utilize globally distributed supply chains in order to take advantage of lower labor costs in different countries. Changes in both the value and the volatility of labor costs can affect the production costs unexpectedly. At the same time, the product architecture i.e. the way product is assembled, and the quality and cost of components can be major factors affecting the production cost. This study proposes a stochastic optimization model for deciding the optimal product architecture and the supply chain configuration simultaneously, considering randomness in labor costs. Simultaneous consideration of these decisions enables improvements in product design as well as supply chain. 3 - The Impact of the Supply Chain on the Success of Product Pre- announcements Laharish Guntuka, Doctoral Student, University of Maryland College Park, 3346 Van Munching Hall, College Park, MD, 20742, United States, lguntuka@rhsmith.umd.edu, Curt Grimm, David E. Cantor We explore factors that influence a consumer’s willingness to purchase a competitor’s pre-announced product. Using competitive dynamics and relational view theories, we investigate how issues in the supply chain moderate these relationships. Example of factors explored include competitive aggressiveness, geographic and reputational concerns. We designed a behavioral experiment to test our research questions. TC35

351C Data-driven Analysis for Air Transport Sponsored: Aviation Applications Sponsored Session Chair: Catherine Cleophas, RWTH Aachen University, Aachen, 52072, Germany, catherine.cleophas@rwth-aachen.de Co-Chair: Virginie Lurkin, Ecole Polytechnique Fédérale de Lausanne, Route Cantonale, Lausanne, 1015, Switzerland, virginie.lurkin@epfl.ch 1 - The Impact of Airline Market Competition Structure on Fares Susan Hotle, Assistant Professor, Virginia Tech, 200 Patton Hall, 750 Drillfield Drive, Blacksburg, VA, 24061, United States, shotle3@vt.edu, Stephanie Atallah In recent years, changing economic conditions have led airlines to realign service to the most profitable routes and reduce operating costs. This presentation evaluates how non-stop airline markets have evolved with an emphasis on the continuity of service and the relationships between competition structure, market airport sizes, and fares. This analysis is performed using DB1B, OAG airline schedules, and the FAA’s Airport Classification on size for years 2005-2015. 2 - A Nonparametric Approach to Estimating Air Travel Demand Structures from Panel Data Analyzing historical sales data to draw conclusions on the underlying demand structure is a central theme of revenue management. This contribution focuses on the estimation of demand segments present in a market using nonparametric methods on panel data. We employ finite mixtures to model purchase events over time frames and obtain estimators for purchase probabilities and market share. The approach shows promising performance when applied to simulated panel data sets with a MNL customer choice model. 3 - Using Online Surveys to Develop Choice Set Generation Models Laurie A.Garrow, Georgia Institute of Technology, School of Civil & Environmental Engr, 790 Atlantic Drive, Atlanta, GA, 30332-0355, United States, laurie.garrow@ce.gatech.edu, Ziran Chen, Mohammad Ilbeigi We designed an online experiment to capture how individuals filter and screen alternatives for discrete choice modeling applications. We ran this experiment on Amazon Mechanical Turk (AMT) and Qualtrics. We present preliminary results comparing responses obtained from AMT (a crowdsourcing Internet marketplace), Qualtrics (a traditional marketing survey panel), and an online survey conducted by the Resource Systems Group that recruited participants from airlines’ frequent flyer databases. Based on these results, we offer a conceptual framework for how online surveys can be used to develop a choice set generation model for airline itinerary choice applications. Johannes Jörg, RWTH Aachen University, Aachen, 52072, Germany, johannes.ferdinand.joerg@ada.rwth-aachen.de, Catherine Cleophas

TC34

351D Operations Research and Intelligence Sponsored: Military Applications Sponsored Session

Chair: Andrew Oscar Hall, United States Military Academy, West Point, NY, 10996, United States, andrew.hall@usma.edu 1 - Hazardous Materials (hazmat) Transport Risk Assessment Considering Terrorist Assaults Zafer Yilmaz, McGill University, Montreal, QC, Canada, zyilmaz1996@gmail.com, Vedat Verter Terrorist attacks against military convoys including hazmat trucks have increased during the past decade. We developed a methodology that incorporates the risk of such attacks as well as the accident risks. A case study focusing on the eastern part of Turkey suffering from terrorist attacks will be presented. Based on data from the past five years, the proposed risk assessment method is incorporated in a GIS framework so as to identify the best paths among the common origin- destination pairs.

348