2015 Informs Annual Meeting

POSTER SESSION

INFORMS Philadelphia – 2015

4 - Analysis of Best Practices for Energy Efficient Buildings through Building Energy Modeling in Design Chung-suk Cho, Assistant Professor, Khalifa University, Al Saada St. and Muroor Rd., Abu Dhabi, 127788, United Arab Emirates, chung.cho@kustar.ac.ae, Young-ji Byon Building energy performance modeling evaluates energy efficient design options. There are significant amount of misused opportunities of energy efficiency-related decisions that could be incorporated into the building design through quality building energy performance modeling. The best practice analysis will help optimize the building design and will allow the design team to prioritize investment in the strategies that will have the greatest effect on the building’s energy use. 5 - Two-stage Optimal Demand Response with Battery Energy Storage Systems Yanyi He, Senior Scientist, IBM, 1001 E Hillsdale Blvd, Foster City, CA, 94404, United States of America, heyanyidaodao@gmail.com, Zhaoyu Wang Proposes a two-stage co-optimization framework for the planning and energy management of a customer with battery energy storage systems (BESSs) and demand response (DR) programs. The first stage is to assist a customer to select multiple DR programs to participate and install batteries to coordinate with the demand side management. The second stage is to perform energy management according the planning decisions, including dispatches of batteries, loads, and DGs. Software Demonstration Cluster: Software Demonstrations Invited Session 1 - Statistics.com - A Survey of Data Analytics Methods Peter Bruce, Founder and President, The Institute for Statistics Education at Statistics.com This workshop will survey the field of data analytics, reviewing both traditional statistical methods and machine learning methods, including predictive modeling, unsupervised learning, text mining, statistical inference, time series forecasting, recommender systems, network analytics, and more. It will be a broad brush treatment aimed at newcomers, as well as those with knowledge in one area who wish to understand where other analytic methods fit into the picture. 2 - River Logic - Code-free modeling for large-scale LP and MIP problems using Enterprise Optimizer Eric Kelso, VP Product Management, River Logic Enterprise Optimizer is a code-free, visual LP and MIP optimization modeling platform. Using EO’s intuitive drag-and-drop interface, learn how to rapidly create integrated process and financial models. Also learn about EO’s wizard- driven data integration, query designer, user-defined schema, dashboard builder, VBA integration, APIs and job automation component. Outputs demonstrated include detailed unit costs and audit-quality P&L, Balance Sheet and Cash Flow statements. The entire session will be spent discussing major features and showing real-world applications. Chair: Min Wang, Drexel University, 3141 Chestnut Street, Philadelphia, PA, United States of America, mw638@drexel.edu Co-Chair: Allen Holder, Rose-Hulman Mathematics, Terre Haute, IN, United States of America, holder@rose-hulman.edu Co-Chair: Wenjing Shen, Drexel University, Philadelphia, PA, United States of America, ws84@drexel.edu 1 - Surgery Scheduling with Recovery Resources Maya Bam, University of Michigan, Industrial and Operations Engineering, 1205 Beal Ave., Ann Arbor, MI, 48109, United States of America, mbam@umich.edu, Mark Van Oyen, Mark Cowen, Brian Denton Surgery scheduling is complicated by the post-anesthesia care unit, the typical recovery resource. Based on collaboration with a hospital, we present a novel, fast 2-phase heuristic that considers both surgery and recovery resources. We show that each phase of the heuristic has a tight provable worst-case performance bound. Moreover, the heuristic performs well compared to optimization based methods when evaluated under uncertainty using a discrete event simulation TB79 79-Room 302, CC Tuesday, 12:30pm - 2:30pm Exhibit Hall A Tuesday Poster Session Contributed Session

model. 2 - Inventory Control with Unknown Demand and Nonperishable Product

Tingting Zhou, Rutgers University, 1 Washington Park, Newark, NJ, 07102, United States of America, tingzhou@rutgers.edu, Michael Katehakis, Jian Yang We study an inventory control problem with unknown discrete demand distribution, focusing on the analysis of an adaptive algorithm based on empirical distributions and the newsvendor formula. When items are nonperishable, the algorithm can achieve a near square-root-of-T bound on its regret over the ideal case where demand distribution were known. 3 - Optimizing Information System Security Investments with Risk: Insights for Resource Allocation Yueran Zhuo, PhD Candidate, University of Massachusetts Amherst, Isenberg School of Management, Amherst, MA, 01003, United States of America, yzhuo@som.umass.edu, Senay Solak Information security has become an integral component of a firm’s business success, and thus investing on information security countermeasures is an important decision problem for many businesses. We use a portfolio approach to study the optimal investment decisions of a firm, where the uncertainty of information security environment is captured through a stochastic programming framework. Results cast managerial insights for information security investment planning by a firm. 4 - Adaptive Decision-Making of Breast Cancer Mammography United States of America, fxw005@uark.edu, Shengfan Zhang The American Cancer Society currently recommends all U.S. women undergo routine mammography screenings beginning at age 40. However, due to the potential harms associated with screening mammography, such as overdiagnosis and unnecessary work-ups, the best strategy to design an appropriate breast cancer mammography screening schedule remains controversial. This study presents a mammography screening decision model that aims to identify an adaptive screening strategy while considering disadvantages of mammography. We present a two-stage decision framework: (1) age- specific breast cancer risk estimation, and (2) annual mammography screening decision-making based on the estimated risk. The results suggest that the optimal combinations of independent variables used in risk estimation are not the same across age groups. Our optimal decisions outperform the existing mammography screening guidelines in terms of the average loss of life expectancy. While most earlier studies improved the breast cancer screening decisions by offering lifetime screening schedules, our proposed model provides an adaptive screening decision aid by age. Since whether a woman should receive a mammogram is determined based on her breast cancer risk at her current age, our “on-line” screening policy is adaptive to a woman’s latest health status, which causes less bias in reflecting the individual risk of every woman. 5 - Optimization of Netting Scheme in Large-scale Payment Network Shuzhen Chen, University of Science & Technology of China, No. 98, Jinzhai Road, Hefei, China, csz@mail.ustc.edu.cn As netting becomes combined with real-time settlement, an efficient netting method is required to deal with the large-scale payment network. Network optimization may not be optimal due to repeated searching of shortest path. A new method is proposed to optimize the netting process by assembling payments in two specific routes. It can minimize the amount of total payments for the whole network and ensure unchanged net payment for each bank. Moreover, it has polynomial time-complexity. 6 - Wasserstein Metric and the Distributionally Robust TSP Mehdi Behroozi, University of Minnesota, Minneapolis, MN, United States of America, behro040@umn.edu, John Gunnar Carlsson Recent research on the robust and stochastic travelling salesman problem and the vehicle routing problem has seen many di?erent approaches for describing the region of uncertainty, such as taking convex combinations of observed demand vectors or imposing constraints on the moments of the spatial demand distribution. One approach that has been used outside the transportation sector is the use of statistical metrics that describe a distance function between two probability distributions. In this paper, we consider a distributionally robust version of the Euclidean travelling salesman problem in which we compute the worst-case spatial distribution of demand against all distributions whose earth mover’s distance to an observed demand distribution is bounded from above. This constraint allows us to circumvent common overestimation that arises when other procedures are used, such as fixing the center of mass and the covariance matrix of the distribution. Screening: A Heuristic-Based Regression Model Fan Wang, University of Arkansas,, Fayettevlle, AR,

311