Informs Annual Meeting 2017

TE01

INFORMS Houston – 2017

Tuesday, 4:35 - 6:05PM

TE02

310B Decision Analysis, Game Theory, Homeland Security, and Disaster Management, Part II Sponsored: Decision Analysis Sponsored Session Chair: Jun Zhuang, University at Buffalo, SUNY, Buffalo, NY, 14260, United States, jzhuang@buffalo.edu 1 - Mapping Terrorist Networks with Graphical Databases Alexander Gutfraind, Uptake, Inc., 600 W. Chicago Avenue, Chicago, IN, 60654, United States, sasha.gutfraind@uptake.com Open datasets about terrorist networks are scarce, small, and frequently suffer from incomplete data. To address this problem, we developed a system for recording data on terrorist events based on open media sources using graphical databases, and applied it to ISIS networks in Europe. We found robust differences between covert ISIS networks and Al-Qaida’s networks from the earlier generation of Islamist terrorism. We suggest that graphical databases promise to Jose Emmanuel Ramirez-Marquez, Stevens Institute of Technology, Castle Point on Hudson, School of Systems & Enterprises, Hoboken, NJ, 07030, United States, jmarquez@stevens.edu, Gabriela Noemi Gongora Svartzman After the attacks on 9/11 the government of USA took extra effort into tracking potential terrorist threats to their country. Around the world public transportation systems have been a popular target for terrorists. This work shows the effect of a possible attack on the subway system of NYC, measuring alternatives and resilience of the system itself. The stages of resilience measured include the reliability of the system before an attack, the vulnerability and adaptability after an event. The data is drawn from daily crimes reported at subway stations, inflows and outflows, and census data. A discrete event simulation of the subway system is modeled and used to evaluate an attack on the subway. 3 - Decision Analytic Techniques in Practice for Homeland Security Problems Julia A. Phillips, Argonne National Laboratory, Oswego, IL, United States, phillipsj@anl.gov This discussion presents an overview of decision analytic techniques that have been deployed for use in the homeland security and resilience operational space. This dialogue will also cover several tools that are currently in use based off these decision analytic techniques. 4 - Deductible-based Public Assistance Program Allison C. Reilly, Assistant Professor, University of Maryland, College Park, MD, United States, areilly2@umd.edu, Gina Tonn, Seth Guikema, Hamed Ghaedi Recently, FEMA proposed establishing a deductible-based approach for its Public Assistance (PA) Program. The new rule is designed to encourage States to engage in pre-disaster planning and mitigation by offering targeted reductions to their deductible. This change represents a seismic shift in how post-disaster money is allocated and has the potential to change the hazard environment. This talk will review the proposed rule and discuss how States should best respond given different objectives. 5 - Staying Ahead of the Game: Adaptive Robust Optimization for Dynamic Allocation of Threat Screening Resources Sara McCarthy, University of Southern California, Los Angeles, CA, United States, saramarm@usc.edu, Phebe T. Vayanos, Milind Tambe We consider the problem of dynamically allocating screening resources at checkpoints to successfully avert an attack. Previous work in this area operates under the unrealistic assumption that screenee arrival times are perfectly known. We thus propose a novel framework that explicitly accounts for uncertainty in arrival times. We model the problem as a multistage robust optimization problem and propose a tractable solution approach using compact linear decision rules combined with robust reformulation and constraint randomization. We show that our approach is scalable and outperforms previous state of the art in terms of both feasibility and optimality. enable more robust and reproducible analysis of covert networks. 2 - Evaluating the Resilience of Subway Systems under Simulated Attacks

TE01

310A Judgments and Forecasting Sponsored: Decision Analysis Sponsored Session

Chair: Kenneth Nguyen, University of Southern California, 8211 Somers Drive, Anaheim, CA, 92841, United States, hoangdun@usc.edu 1 - Comparing Verbal and Numeric Forecasts: New Findings and Implications Kenneth Nguyen, University of Southern California, 8211 Somers Drive, Anaheim, CA, 92804, United States, hoangdun@usc.edu, Richard S.John This research compares verbal and numeric forecasts of subjective probability estimates. A sample of 118 NFL football experts was recruited to participate in the study. The experts were randomized into one of the two experimental conditions that differ in the modes of the response scale. Verbal responses were later transformed into numbers using a method based on Savage’s conceptualization of subjective probabilities. Results showed that verbal forecasts were not statistically significant from numerical predictions in terms of the overall accuracy. However, numerical judgments were more discriminatory and more biased than verbal judgments. 2 - Quantifying the Accuracy of Subjective Probability Estimates: A Meta-analysis Kenneth Nguyen, University of Southern California, 8211 Somers Drive, Anaheim, CA, 92804, United States, hoangdun@usc.edu, Matthew Baucum, Richard S. John We attempted to address the question whether subjective probability estimates are actually better than random guesses, and explored the effects of various methodological and substantive factors. A comprehensive database search returned 466 records, and 84 of these met the study’s inclusion criteria. Preliminary data analyses suggest that subjective probability estimates were better than chance. In addition, experts’ judgments were significantly better than laypeople’s judgments. Yet, there was remained large heterogeneity in the effect size, suggesting the roles of other moderators beyond expertise. Data analyses are ongoing, and a full report is expected by September. 3 - The One Man Show: the Connection Between Overconfidence and Joint Decision-making Dominik Piehlmaier, Doctoral Student, University of Wisconsin- Madison, 1300 Linden Dr, Madison, WI, 53706, United States, piehlmaier@wisc.edu, Dee Warmath, Cliff Robb Verbal and nonverbal interactions as well as social networks enable humans to profit from the knowledge and skills of others (Harvey and Fischer, 1997). Advice seeking and taking are integral parts of such interactions and help to increase calibration in one’s decision-making process (See et al., 2011). This study explores the connection between shared decision-making and overconfidence. We draw from a dataset of 2,000 investors (45 % female) and find that individuals who include household members and/or financial advisors in their investment decisions are significantly more calibrated in their financial judgments and less overconfident in terms of monetary expectations towards future growth. 4 - Judgement Error in Lottery Play: When the Hot Hand Meets the Gambler’s Fallacy Qingxia Kong, Assistant Professor, Erasmus University, Rotterdam, Netherlands, q.kong@rsm.nl, Chung-Piaw Teo, Nicolas Lambert We use two sets of naturally occurred data to show that both the gambler’s fallacy and hot-hand fallacy can exist in lottery games. The existence of hot-hand fallacy is surprising, as previous works have documented instead the presence of an opposite effect, the gambler’s fallacy, in the US lottery market. Some literature also suggest that gambler’s fallacy prevails when random numbers are generated by mechanical devices such as in lottery games. We show that our empirical findings are consistent with the classical optimal Bayesian learning model, which allows us to investigate some conditions in the lottery game design that determine which phenomenon will dominate.

408