Sunday, April 10
Take advantage of in-depth training in software solutions provided
There is no charge for these
workshops if you register for the INFORMS Conference - sign up
Introduction to Bayesian Networks and Their Applications in Marketing Science
Stefan Conrady, Conrady Applied Science, LLC; Lionel Jouffe, Bayesia SAS
Bayesian networks have emerged as a new form of probabilistic knowledge representation. Over the last 25 years the properties of Bayesian networks have been fully validated in the world of academia and they are now becoming powerful and practical tools for “deep understanding” of very complex, high-dimensional problem domains. This tutorial will provide an introduction to the wide-ranging applications of Bayesian networks in the field of marketing science. Participants do not need to have any prior familiarity with Bayesian networks. Participants will learn how Bayesian networks can serve as innovative alternatives to traditional modeling techniques in market research, from simple regressions to elaborate discrete choice models. The examples in this tutorial will be based on recent, large-scale consumer surveys from the auto industry, demonstrating the practical relevance of Bayesian networks in market research.
Crossing the Chasm: Bringing Decision Framing for Project Management and the Insight Gained from Practical Uncertainty Analysis to the Everyday Organization
Ellen Coopersmith, Decision Frameworks
How do you take a room full of issues and bring clarity of action on key decisions quickly? Through quality decision making – a “three legged stool” comprised of (1) efficient decision framing, (2) practical uncertainty analysis, and (3) timely dialogue with decision makers. This workshop teaches the decision framing and uncertainty analysis workflow through an interactive joint venture case study, and the use of Decision Frameworks DTrio framing and decision analysis workflow software and TreeTop, the easiest-to-use and most powerful decision tree software. To truly “cross the chasm” and make quality decision making integral to corporate culture, the practice of decision framing and uncertainty analysis must leave the hands of a few innovative users” and reside in the workflow and toolset of “many pragmatic users.” This workshop is all about making quality decision making easy.
Optimization Modeling with MPL and Python: Integrating Optimization in Real-World Business Applications
The workshop focuses on optimization modeling and how to embed optimization models into end-user applications. It is designed to teach participants how to develop their own optimization models, connecting them to data sources, and solve the models using state-of-the-art, industrial strength solvers. We will take you through all the steps of implementing O.R. projects using optimization software tools from Maximal Software, IBM-ILOG and Gurobi. Participants will leave with an excellent overview of optimization modeling and a good working knowledge of how to successfully build optimization applications. We will also demonstrate the new "MPL for Python," a scripting and library interface for the MPL Modeling Language.
Risk and Decision Assessment using @RISK
The workshop starts with a brief review of some core principles, best practices and tools involved in the building of good Excel models (such as those relating to model design, structure and format, sensitivity analysis, etc.), before highlighting some of the limitations of such methods, and discussing how Palisade’s tools add value. You will (1) gain some knowledge of key principles and best practices in general Excel modeling, including the ability to recognize where pure Excel-based approaches are insufficient; (2) recognize situations where @RISK could be of use; and (3) gain sufficient knowledge of @RISK to build simple yet powerful models for realistic applications in risk modeling, decision analysis, and optimization.
Introduction to Salford Systems’ Data Mining Suite
Salford Systems offers a highly accurate, ultra-fast predictive and data mining platform for developing models from databases of any size, complexity or organization. Our technologies span classification, regression, survival analysis, missing value analysis and clustering/segmentation. Core components of our platform include CART®, MARS®, TreeNet® and RandomForests®.
Gurobi Performance Marches On
Bob Bixby and Greg Glockner
This workshop will review recent improvements in the Gurobi Optimizer. With Version 4.0, our emphasis was on the new Gurobi QP solvers, including QP simplex, QP barrier and Mixed Integer QP. Version 4.5 is focused on performance and stability improvements in our LP algorithms and our flagship MIP offering. We will talk about some of the important ideas contributing to Version 4.5 and will conclude with a summary of current benchmarks. The workshop will also include presentations devoted to performance tuning of the Gurobi solvers and a discussion of best practices when building applications using the different Gurobi APIs.
ExtendSim Simulation Software
Simulation is a key technology to understanding your company's business processes and identifying opportunities for improvement. Experience the ease at which robust, intuitive models are built with the power tool for simulation - ExtendSim. Use ExtendSim to achieve your modeling objectives rapidly and be assured success. ExtendSim features a customizable graphical modeling logic where relationships between modeling components are visible. Its flexibility provides modelers with the ability to create and couple hierarchical subsystems, create interactive user interfaces, and the freedom to rapidly explore design options. The internal database's user-friendly graphical interface facilitates organization of complex data structures to efficiently test multiple scenarios. This session will include an ExtendSim modeling building session and demonstrate a range of real world models from a variety of industries.
Developing Optimization, Simulation and Business Analytics Applications in Rapid Development Mode
Ravindra K. Ahuja, Innovative Scheduling, Inc.
Innovative Scheduling develops optimization, simulation and business analytics solutions in a rapid development mode. We license software products as well as develop custom software applications in logistics, transportation and scheduling. Using a pre-existing library of optimization and simulations engines, and our framework for map-based, web-based interface development, we can deploy sophisticated applications with attractive user-friendly interfaces in three to six months. In this workshop, we will give an overview of some of our recent optimization, simulation and business analytics projects. We will describe how we approached these business problems, how we modeled them and solved them, how we packaged our approaches into interactive decision support systems, and how we created substantial value for clients. We will share with the audience the do’s, don’ts and the lessons learned.
Get Started Quickly with Predictive Analytics and Data Mining: XLMiner
If you want to move your organization from simple reporting and statistics to extracting insights from your business data through predictive analytics, you have a choice: invest lots of money and time in an enterprise tool with a steep learning curve, or start mining your data today, at nominal cost, in Microsoft Excel with XLMiner. It’s an easy choice if you want to deliver quick results, but the surprise may be how far you can go. This workshop will show you XLMiner-based data preparation and clustering, classification and prediction tools, using your choice of neural nets, classification and regression trees, naïve Bayes, logistic regression, k-nearest neighbors, discriminant analysis, principal components, time series forecasting, and association rules. Get started quickly with a free trial software CD.
An Overview of the KNITRO Nonlinear Optimization Solver
This workshop will present an overview of the features and usage of the KNITRO nonlinear optimization solver. Specifically the workshop will cover the following topics: (1) a very brief overview and background on nonlinear optimization will be provided; (2) a description of the KNITRO software and its main features will be presented; (3) examples will be shown of how to use KNITRO as an embeddable callable library and through various modeling languages; and (4) case studies will be presented demonstrating the power of KNITRO in particular application areas.
Optimization and Simulation Models from Excel to your Intranet or Web Server
Bring your organization up the analytics learning curve quickly, and get an early payoff with easy model-building in Excel, yet ultra-high performance that once required custom development. See how easily you can go from a prototype spreadsheet model to a full-scale application on your corporate server – in just minutes with Risk Solver Platform and Solver SDK Platform. We’ll take you step by step from creating a model in Excel to solving that model directly in a multi-core application using C++, C# or Java, on Windows, Linux or Azure. Solve conventional LP/MIP models, nonlinear and non-smooth models, Monte Carlo simulation models, and use robust optimization, stochastic programming and simulation optimization. Learn what you can do, and take away a free trial software CD to do it!
Applying Analytics and Optimization to the Supply Chain
Michael Watson, IBM ILOG Supply Chain Executive
This workshop will cover how researchers and practitioners in supply chain have long delivered analytics and optimization solutions to their clients. These solutions have ranged from building custom optimization problems to using an off-the-shelf optimization-based product. We have found that custom optimization and off-the-shelf solutions can complement each other and make the practitioner more productive. This talk will provide historical perspective on these solutions up to a recent study showing how leading supply chains are using analytics and optimization. We will rely on case studies and examples throughout the presentation.
How to Boost Solver Performance on Tough Optimization Models
Recent dramatic improvements in the ability to solve optimization models with linear and/or quadratic constraints, and integer variables provide both opportunity and challenge. The challenge is to formulate your model so that it fits into the structure of these easily solved problems. The new versions of LINGO, What’sBest! and LINDO API announced at this meeting have extensive features to automatically reformulate your model so that it fits into these easily solved classes. We will describe modeling tricks and rules to help you make sure that your model is easily solved. We will give examples based on applications in finance, production planning, supply chain management, engineering design, planning under uncertainty, and more.
Solving Business Problems with SAS® Analytics and OPTMODEL
SAS provides a full range of analytic methods, including data integration, statistical analysis, data and text mining, and forecasting, all integrated with our OPTMODEL optimization modeling and solution capabilities. Atop this analytic foundation, OPTMODEL delivers unified access to a wide range of optimization models, including LP, MILP, QP and general NLP. OPTMODEL supports both standard and customized optimization algorithms. In our workshop, we will highlight OPTMODEL’s power and versatility in modeling and in creating solution algorithms. We will also emphasize its deep integration with SAS data, analytic and reporting capabilities, including the creation of business-level user interfaces. We’ll touch briefly on classical models, but will spend most of our time on case studies drawn from our successful work with customers across a broad range of industries.