2016 INFORMS Annual Meeting Program

POSTER SESSION

INFORMS Nashville – 2016

Application Of Multicriteria Methodology For Decision Aid In The Formation Of A Projects Portfolio Marcos Santos, Liutenant Commander, Brazilian Navy, Arsenal da Marinha no Rio de Janeiro (AMRJ), Rua da Ponte S/N, Ilha da Cobras, Centro, Rio de Janeiro, 20091000, Brazil, marcosdossantos_doutorado_uff@yahoo.com.br, Hudson Souza, Fabrício Costa Dias, Ernesto Rademaker Martins, Marcone Freitas Reis Decide correctly is a constant challenge faced by man since the beginning of time. Among the numerous multicriteria methods of decision support, it was used the Analytic Hierarchy Process (AHP), which, as a compensatory method, it seems appropriate to solve such problems. The AHP was one of the first methods developed by the American School, one of the most used methods in the world. This paper aims to propose through AHP, a methodology for constitution the portfolio of IT projects of a non-profit company. Based on this interview it was possible to raise the projects evaluation criteria as well as the preference of the decision maker. Pro Bono Analytics - Informs Volunteers Create Societal Impact With Applied Analytics Rina R Schneur, ARCKS, Lexington, MA, 02420, United States, rinarsg@gmail.com, Michael P Johnson Pro Bono Analytics was established by INFORMS in 2015, in the tradition of other disciplines’ efforts to utilize specialized skills and knowledge to generate social impact. Pro Bono Analytics’ goal is to provide analytics technical support for nonprofit organizations without the capacity and/or resources to perform data analysis related tasks on their own. This poster presentation will provide knowledge about why INFORMS members should consider volunteering for Pro Bono Analytics, how this initiative works, and what promising current and recently-completed engagements look like. MLB And Regression Analysis. Predictions For The 2016 Chicago Cubs And White Sox Logistic and linear regression models were built to predict outcomes for the 2016 Chicago Cubs and White Sox. The models in concert predicted overall wins, runs scored, runs allowed, and finally the predicted playoff status for each team. Future directions in web scraping and model building are discussed. Fuzzy-logic / Dempster - Shafer Based Information Fusion Formulism For Land-marine Decision Analysis Nicholas V Scott, Spectral Scientist/Physical Oceanographer, Riverside Research, 2640 Hibiscus Way, Beavercreek, OH, 45431, United States, nscott@riversideresearch.org A land-marine problem is heuristically addressed using a fuzzy logic/Dempster- Shafer based information fusion formulism which demonstrates the efficacy of such tools as aids in optimal decision making. The initial computational segment contains a five component feature extraction system which provides the inputs to a fuzzy logic inference system. Multiple human assessments, which emanate from the use of the inference system and ancillary intelligence, are then amalgamated using Dempster-Shafer evidential theory. A probabilistic assessment of environmental state is provided finally allowing for decisions in which information ignorance and data uncertainty are taken into account. Reducing Social Risks In The Supply Chain: An Examination Of S&P 500 Companies Approximately half of S&P 500 companies report implementing initiatives to reduce social risks in the supply chain. Based on Bloomberg data, these S&P 500 companies are compared to those without such initiatives in terms of firm characteristics (e.g., size, industry sector), related policies (e.g., child labor, human rights, environmental) and profitability (e.g., return on assets). Optimal Balanced Sample Selection For Causal Inference Using Machine Learning Dhruv Sharma, Graduate Student, George Washington University, Washington, DC, 20429, United States, dhruvsharma@gwmail.gwu.edu With the availability of observational survey data and big data the ability to sample accurately to determine causal effects beyond correlational studies is important. This paper investigates machine learning supervised ensemble classification Area Under the Curve (AUC) measure, for optimization of balanced sample selection. Synthetic data sets and actual experimental data are used to compare results of existing optimization metrics. Rose Sebastianelli, Professor, University of Scranton, Brennan Hall 423, Scranton, PA, 18510, United States, rose.sebastianelli@scranton.edu, Nabil Tamimi Kurt J Schuepfer, Graduate Researcher, Miami University, Oxford, OH, 45056, United States, schuepferk@gmail.com

Adaptive Sampling Trust Region Algorithms For Derivative Free Simulation Optimization Sara Shashaani, Purdue University, 782 N Commodores Ln., Lafayette, IN, 47909, United States, sshashaa@purdue.edu, Raghu Pasupathy We develop derivative free algorithms for optimization contexts where the objective function is observable only through a stochastic simulation. The algorithms we develop follows the trust-region framework where a local model is constructed, optimized, and updated as the iterates evolve through the search space. The salient feature of our algorithms is the incorporation of adaptive sampling to keep the quality of the local model in lock step with the trust-region radius, in a bid to ensure optimal convergence rates. Ruled Based Prediction Analysis For 30-days Neurological Recovery Status Post Stand Assisted Treatment Of Brain Aneurysm Karmel Shehadeh, PhD Student, University of Michigan, 1205 Beal Avenue, Ann Arbor, MI, 48109, United States, ksheha@umich.edu, Chun An-Chou Recently, it has been observed that stroke patients could recover with asymptomatic outcome in a short period with use of stent-assisted coiling (SAC) treatment. We employed a rule-based decision model to identify key rules that are used for predicting the clinical outcomes post 30-Days of SAC treatment. A 95% and 75% prediction accuracy were obtained for a cohort of 65 training and 21 validation patients, respectively. The Impact Of Social Feedback On Reviewers’ Review Decisions Wenqi Shen, Virginia Tech, Blacksburg, VA, United States, shenw@vt.edu, Yan Liu In this paper, we empirically examine how social incentives, namely online reputation and social feedback which reflects peer recognition and attention, affect reviewers’ review decisions. We develop a state-space model which captures the dynamics of reviewers’ incentives as influenced by both online reputation and social feedback. Quay Crane Scheduling Problem With Considering Tidal Impact And Fuel Consumption Yu Shucheng, doctor, Shanghai university, Shang Da Road 99, Shanghai 200444, China, Shanghai, 200444, China, yushucheng2007@163.com This study investigates a quay crane scheduling problem with considering the impact of tides in a port and fuel consumptions of ships. A mixed-integer nonlinear programming model is proposed. Some nonlinear parts in the model are linearized by approximation approaches. For solving the proposed model in large-scale problem instances, both a local branching based solution method and a particle swarm optimization based solution method are developed. Numerical experiments with some real-world like cases are conducted to validate the effectiveness of the proposed model and the efficiency of the proposed solution methods. A Dynamic Programming Approach To Solve Bi-level Programming Problem With Fuzzy Rule-base Constraints Vishnu Pratap Singh, Research Scholar, Indian Institute of Technology-Kharagpur, Department of Mathematics, Kharagpur, WB, 721302, India, vishnupratapsingh56@gmail.com In this work, A bi-level programming problem has been considered where the functional relationship between decision variables and the objective functions of leader and follower are not completely known to us. So a bi-level programming problem with fuzzy rule-base constraints has been developed. A dynamic programming approach with appropriate fuzzy reasoning scheme is used to determine the crisp functional relationship between the objective functions and the decision variables. Thus a bi-level programming problem is formulated from the original fuzzy rule-based to the conventional bi-level programming problem. Using Discrete Event Simulation To Improve Acute Stroke Care Quality Measurement Lina Song, PhD Candidate, Harvard University, 14 Story Street, 4th floor, Cambridge, MA, 02138, United States, dahye.lina.song@gmail.com Time from stroke onset to the administration of tissue plasminogen activator (tPA) is an important acute stroke care performance measure, but it should be adjusted for the operational characteristics of hospitals to avoid setting unrealistic benchmarks for smaller hospitals. We developed a discrete event simulation model to compare the time-to-tPA among four types of hospitals with varying stroke-related resources. Stroke patients arrive at an emergency department (ED) according to a Poisson process and navigates through the system. According to the model, larger comprehensive stroke centers can achieve better performance on time-to-tPA measures compared to non-stroke centers.

296