2016 INFORMS Annual Meeting Program

MD79

INFORMS Nashville – 2016

MD94 5th Avenue Lobby-MCC Technology Tutorial: Gurobi/AMPL Technology Tutorial 1 - Gurobi Optimization Dr. Greg Glockner, Vice President of Engineering, Bellevue, WA, glockner@gurobi.com Advanced Python Modeling with Gurobi Are you looking for an environment that combines the expressiveness of a modeling language with the power and flexibility of a programming language? The Gurobi Python interface allows you to build concise and efficient optimization a model using high-level modeling constructs. This tutorial will provide an overview of these capabilities, including an introduction to new modeling features that significantly enhance the expressiveness of our environment. 2 - AMPL In The Cloud: Using Online Services To Develop And Deploy Optimization Applications Through Algebraic Modeling Robert Fourer, AMPL Optimization Inc., 2521 Asbury Ave, Evanston, IL, 60201, United States, 4er@ampl.com Optimization modeling systems first appeared online almost 20 years ago, not long after web browsers came into widespread use. This presentation describes the evolution of optimization alternatives in what has come to be known as cloud computing, with emphasis on the role of the AMPL modeling language in making models easy to develop and deploy. We start with the pioneering free NEOS Server, and then compare more recent commercial offerings such as Gurobi Instant Cloud; the benefits of these solver services are readily leveraged through their use with the AMPL modeling tools. We conclude by introducing QuanDec, which creates web-based collaborative applications from an AMPL models. Robert Fourer, an authority on the design and implementation of computer software to support large-scale optimization, studied at M.I.T. and Stanford and was a professor of Industrial Engineering and Management Sciences for over 30 years. He is a founder and is currently President of AMPL Optimization Inc. and is co- author of a popular book on modeling in the AMPL language. Machine Learning II Sponsored: Data Mining Sponsored Session Chair: Hamidreza Ahady Dolatsara, MD, United States, hamid@auburn.edu 1 - Expectation Maximization For Finite Mixture Of Linear Regression Models With Group Structure Data Haidar Almohri, PhD Candidate, Wayne State University, 2833 Catalpa Circle, Ann Arbor, MI, 48108, United States, almohrih@yahoo.com, Ratna Babu Chinnam, Arash Ali Amini One of the limitations with the available methods for fitting Finite Mixture of Linear Regression (FMR) models is that they do not explicitly account for group structure within the datasets during modeling. It is sometimes desirable to force the algorithm to allocate groups/blocks of observations to individual models, instead of individual observations. We propose an Expectation Maximization (EM) approach to fit FMR to data with group structure. 2 - A Comparison Method For Association Rule Mining Algorithms Gulser Koksal, Prof. Dr., Middle East Technical University, Industrial Engineering Department, Ankara, 06800, Turkey, koksal@metu.edu.tr, Sanam Azadiamin Association rule mining algorithms are helpful in extracting useful information from large amount of data. In literature, these algorithms are compared using data sets for which interesting association rules are unknown. A novel comparison method is developed to perform the comparison using designed data sets containing interesting rules generated by logistic regression. Several comparison measures are defined. Statistical analyses and multi-criteria decision making approaches are applied to find the best algorithm considering the selected measures. Tuesday, 8:00AM - 9:30AM TA01 101A-MCC

3 - The Impact Of Manufactures Comptition To Pay Incentive Funds On Supply Chain Neda Khanjari, Assistant Professor, Rutgers University, 227 Penn St, BSB 260, Camden, PA, 08102, United States, neda.khanjari@rutgers.edu In many industries manufacturers pay incentive funds to retailer hired sales agents to boost the demand for the manufacturers’ products. In this paper, we study a supply chain in which two manufactures are competing to get the attention of the retailer hired sales agent to boost their own product. 4 - Partial Outsourcing And Linked Learning Processes Burcu Tan, Tulane University, A. B. Freeman School Of Business, Firms are increasingly outsourcing software and technology development as well as other knowledge work. Standard economic models predict that firms should outsource either all or none of a particular activity; however, recent evidence contradicts that. We develop a dynamic optimization model to provide a rational explanation for partial outsourcing. Our explanation hinges upon the linked learning processes at the subsystem development level and the systems integration level. 5 - A Hybrid Stochastic Fuzzy Approach For Inventory Control In Multi Item Job Shop Processes With Unsteady Lead Time Alcides Santander, Universidad del Norte, Barranquilla, Colombia, asantand@uninorte.edu.co, Kevin Melendez, Nathalia Hernandez, Diego Guillen, Marco E. Sanjuan Uncertainties inherent in customer demands and variable lead times make difficult for supply chains to achieve either just-in-time inventory replenishment or economic optimal process. This results in a low service level among the different echelons of the supply chain. A hybrid stochastic-fuzzy approach for inventory control is proposed in order to guarantee production rate despite fluctuations in demand and process lead time to minimize the work in process. Demand forecast and probabilistic depiction of lead times, along with the process variables, are the inputs of a fuzzy inference system designed to determine whether it is profitable to generate a purchase order in a given period of time. 7 Mcalister Dr., New Orleans, LA, 70118, United States, btan@tulane.edu, Edward Anderson, Geoffrey Parker, Xiaoyue Jiang Utku Tarik Bilgic, Middle East Technical University, Department of Industrial Engineering, Middle East Technical University, Ankara, 06800, Turkey, utbilgic@metu.edu.tr, Sakine Batun Single-period appointment scheduling for a given sequence of outpatient procedures is a well-studied problem. We consider the problem determining the sequence of and the appointment times for a given set of surgeries to be operated over a multi-period planning horizon in the presence of patient no-shows and uncertainty in surgery durations. We formulate and solve the problem as a two- stage stochastic program to estimate the value of capturing uncertainty. 2 - Robust Design Of A Stroke Hospital Network Amir Ardestani Jaafari, McGill University, 1001 Rue Sherbrooke West, Montreal, QC, H3A 1G5, Canada, amir.ardestani-jaafari@hec.ca, Beste Kucukyazici With advances in the diagnosis and treatment of acute stroke, timely medical intention is increasingly critical; however, simply transporting the patients to the closest stroke hospital may cause congestions in some hospitals, while the stroke beds are underutilized in other hospitals. In this research, we study the trade-off between minimizing transportation time to the hospital and minimizing the congestion experiences by the patient at the hospital using robust optimization. 3 - Method To Assign Specialties To Timetable Slots In Surgery Units To Smooth Postoperative Inward Bed Demand Flavio S Fogliatto, Prof., Federal Univ of Rio Grande do Sul, Av. Osvaldo Aranha, 99/5o andar, Porto Alegre, RS, 90040020, Brazil, ffogliatto@producao.ufrgs.br, Marcos Gerchman, Jeruza Neyeloff, Michel J Anzanello, Beatriz D Schaan Peaks in patients’ demand for inward hospitalization usually lead to disruptions in the provision of healthcare, having negative effects on patient and staff satisfaction. A main source of inward bed demand is the surgical theater. This paper proposes a method to determine the best assignment of specialties to timetable slots in surgical theaters, such that the variance of inward bed demand is minimized. For that, we use integer programming heuristics. Our propositions are tested in the surgical unit of a large public University hospital located in the south of Brazil. We were able to reduce the inward bed demand variability by 90%, smoothing the flow of post-surgical patients to hospital wards. MD79 Legends G- Omni Health Care, Modeling VIII Contributed Session 1 - Multi-period Appointment Scheduling In Outpatient Procedure Centers

232