2016 INFORMS Annual Meeting Program

TA22

INFORMS Nashville – 2016

TA19 106B-MCC Optimization Modeling and Beyond with a Focus on Practice Sponsored: Computing Sponsored Session Chair: Leo Lopes, SAS, SAS Campus Drive, Cary, NC, 27513, United States, Leo.Lopes@sas.com 1 - Optimization Modeling With Python Using Pandas Irv Lustig, Princeton Consulting, irv@princeton.com The Python library pandas (http://pandas.pydata.org/) is a popular library used by data scientists to carry out an entire data analysis workflow in Python. When building optimization models, we often work with data in tables that are sourced from databases, CSV files, and spreadsheets. pandas provides a uniform environment for working with data tables with a large number of methods for manipulating tabular data, many of which are directly applicable for building large scale optimization models. In this talk, we will illustrate some of these powerful features that can accelerate optimization model development and deployment. 2 - SAS/OR Value Beyond the Model Leo Lopes, SAS Institute, Leo.Lopes@sas.com We focus on uses of SAS/OR that go beyond modeling and solving, but are are just as essential to deliver production quality prescriptive analytics models quick- ly. The tasks we describe support testing, instance generation, access to alterna- tive solvers, data manipulation, algorithmic control, and visualization. 3 - Building Optimization-enabled Applications Using AMPL Api Robert Fourer, AMPL Optimization Inc., 2521 Asbury Ave, Evanston, IL, 60201, United States, 4er@ampl.com We describe how to combine the power of the AMPL modeling system and a general-purpose programming language to build rich optimization-enabled client applications. Having an optimization model expressed in a high-level declarative form with model and data separation facilitates its evolution and maintenance, and makes switching between different solvers and data sources easy. At the same time it is possible to use a familiar development environment and have access to a wide variety of programming libraries for data management and interface development. 4 - Decision Optimizer 7.0: Combinatorial Optimization For Business Analysts Susanne Heipcke, FICO, San Jose, CA, United States, SusanneHeipcke@fico.com, Livio Bertacco, Sebastien Lannez With Decision Optimizer a strategy analyst can define and optimize complex decision problems using an intuitive graphical workflow. By designing the interactions between decisions and constrained metrics, it is possible to create models for optimizing the assignment of actions, such as investment options or transaction authorization, for large-scale datasets of input elements leveraging analytic techniques for sampling and segmentation. In the new version 7.0, DO has been integrated with FICO Optimization Modeler to provide a more collaborative, web and cloud based experience, improved scenario management, distributed execution and Tableau based reporting. TA20 106C-MCC Mathematical Finance, Models, Simulation and Today’s Pressing Problem Chair: Joseph M. Pimbley, Maxwell Consulting, LLC, 1, Croton-on- Hudson, NY, 10520, United States, pimbley@maxwell-consulting.com 1 - Mathematical Finance, Models, Simulation And Today’s Pressing Problem Joseph M. Pimbley, Maxwell Consulting, LLC, a, Croton-on- Financial markets are awash in information ranging in form from numerical data to unstructured news reports to nebulous narratives of executives and regulators. Investors, fiduciaries, intermediaries and other “market actors” apply an exceedingly broad spectrum of human skill and ingenuity to the interpretation of this streaming information. Mathematical techniques and analysis, in particular, are notable tools in which mathematical advances and discoveries may improve markets’ liquidity, efficiency and pace. This article outlines the origin and techniques of mathematical finance and associated models and simulations. We note strengths and shortcomings of these mathematical tools. The greatest challenge today is to learn and teach to the financial world the necessary judgment to avoid and rescind destructive deployment of financial models. Hudson, NY, 10520, United States, pimbley@maxwell-consulting.com Invited: Tutorial Invited Session

TA21 107A-MCC Beyond Predictive Analytics Sponsored: Health Applications Sponsored Session Chair: Margret Bjarnadottir, University of Maryland, 4324 Van Munching Hall, College Park, MD, 20742, margret@rhsmith.umd.edu 1 - Data-driven Specification Of Cyclical Arrival Processes Donald Lee, Yale University, 165 Whitney Ave, Box 208200, New Haven, CT, 06520, United States, donald.lee@yale.edu, Ningyuan Chen, Sahand Negahban The arrival processes of real-world systems usually exhibit cyclical behaviour. For example, patient arrivals to emergency departments often peak around midday and drops off at night. In this talk we show how this periodic structure can be exploited to obtain a compact and analytic description of the underlying arrival process from data. Such a model is clearly useful for both simulation and modeling purposes. We demonstrate the method on arrivals data from an Emergency Department in southern Connecticut. 2 - Incorporating Dose-prediction Within A Personalized Treatment Paradigm Eva Lee, Georgia Institute of Technology, eva.lee@gatech.edu, Xin Wei This work is joint with Grady Health Systems and Atlanta VA Medical Center. We design an outcome-based decision support tool that couples a predictive treatment-effect model with a planning optimization model. The predictive model uncovers treatment effect analysis of anti-diabetic drug dosage and the blood glucose level recorded in the titration period of each patient. This evidence is then incorporated within a personalized planning model for optimal treatment design. The decision support tool allows continuous learning of evidence for each patient as new treatment outcomes are recorded. 3 - Optimal Selection Of Health Care Providers Jerry Kung, MIT, jkung@mit.edu Given electronic health claims data for employees of a company, we propose a mixed integer optimization approach to select a collection of providers that optimizes over total cost, while maintaining quality and respecting travel distance for employees. We demonstrate that our formulation is tractable for large datasets and present the computational results on a real claims dataset. By following the prescriptions generated by our optimization model, we estimate that cost reductions of up to 10% can be achieved by reassigning patients for a small number of different types of procedures. We demonstrate that these cost reductions are robust to changes in a variety of parameters. TA22 107B-MCC Joint Session MSOM-HC/HAS: Modeling and Optimization for Chronic and End-Stage Renal Disease Patients Sponsored: Health Applications Sponsored Session Chair: Murat Kurt, Merck, Merck, Philadelphia, PA, 07033, United States, murat.kurt7@gmail.com Co-Chair: David Kaufman, University of Michigan-Dearborn, 19000 Hubbard Dr, Dearborn, MI, 48126, United States, davidlk@umich.edu 1 - Optimal Decision Making In A Markov Model With Parameter Uncertainty: The Case Of Chronic Kidney Disease

Reza Skandari, University of British Columbia, Reza.Skandari@sauder.ubc.ca, Steven Shechter

We investigate a Markov decision process whose unknown transition parameters are revealed partially through state observation. Decisions are made as the state evolves. We use the model to study the optimal time to start preparing a type of vascular access for chronic kidney disease patients who will need dialysis.

239