abstracts

Keynote speakers Monday Oct 15, 9 a.m. to noon:

 

Making Models Matter More: From Analytic Excellence to Operational Influence

Michael Schrage
Researcher and Author of Serious Play: How the World’s Best Companies Simulate to Innovate, MIT

The purpose of computing is insight, not numbers. But converting statistically compelling “insights” into “influence” that changes minds and changes behavior is hard. Increasingly, organizational challenges are more difficult for modelers to surmount than technical challenges. Drawing upon extensive real-world research and practice, this talk will offer usable frameworks for how models can be platforms for persuasion as well as tools for analysis.

 

Advanced Systems Design at Georgia Tech’s ASDL

Dimitri Mavris
Boeing Professor of Advanced Aerospace Systems Analysis, Guggenheim School of Aerospace Engineering, Georgia Institute of Technology

The Aerospace Systems Design Laboratory creates unique and innovative methodologies for the design and assessment of technologies of complex vehicles, systems, and system-of-systems. ASDL’s portfolio of methodologies and applications includes formulation, development and implementation of comprehensive approaches to the design of affordable and high-quality complex systems.

The methodologies developed at ASDL enable communication between the technical details and program management structure by speeding up design cycle time through the use of surrogate modeling and probabilistic assessment of assumptions. Technologists present the engineering data in a format that is communicable to specialists in other areas of expertise, project managers and decision-makers. In his presentation, Mavris will demonstrate the enabling capabilities built into JMP to facilitate the mode of communication by showing examples that have been created in conjunction with ASDL activities.

 

Keynote speakers Tuesday, Oct 16 9 a.m. to noon

 

Dynamic Graphics Drive Discovery, and Discovery Drives Innovation

John Sall
Co-Founder and Executive Vice President, SAS

Innovation does not arise from blind creativity. Visualizing data gives us eyes on problems and opportunities. We see data from products and processes, data from experiments, data from needs and desires, and even data that we make from computer models. This is the way we see forward to new solutions. One key to innovation is to put computer visualization techniques to work in the richest and easiest ways. The result helps us see more things and gives us more insight into problems and opportunities. There are many new ways to harness the computer and its graphics power that extend our investigative skills.

 

Data Mining in the Real World: Five Lessons Learned in the Pit

Dick De Veaux
Professor, Department of Mathematics and Statistics, Williams College

Data mining has been defined as a process that uses a variety of data analysis and modeling techniques to discover patterns and relationships in data that may be used to make accurate predictions and decision. Isn’t this what statistics does? Are the two really different? Through a series of case studies, we will try to illuminate some of the challenges of data mining and highlight some of the differences between data mining and traditional statistical analysis. We’ll also show how to avoid the major pitfalls as you embark on your own data mining project.

 

Design and Analysis of Computer Experiments: A New Approach

Bradley Jones
Senior Manager for Statistical Research and Development, SAS

In an effort to speed the development of new products and processes, many companies are turning to computer simulations to avoid the expense and lost time of building prototypes. These computer simulations are often very complex, and it may take hours to complete a single run. If there are many variables affecting the results of the simulation, then it makes sense to design an experiment to gain the most information possible from a limited number of computer simulation runs. The absence of noise is the key difference between computer simulation experiments and experiments in the real world. Since there is no variability in the results of computer experiments, optimal designs based reducing variance have questionable utility. Replication, usually a “good thing,” is clearly undesirable in computer experiments. Thus, a new approach to experimentation is necessary.

JMP® 7 introduces new designs and also a new fitting method specifically created to address the unique behavior of computer simulation models. This talk takes a case study approach using computer models to demonstrate both the new computer simulation design and analysis features in JMP 7.

 

Concurrent Sessions

 

Developing the Corporation’s Analytics Road Map

Tim Pletcher
Director of Applied Research, Central Michigan University Research Corp.

This presentation shares some of the wisdom gleaned from the work to develop a BI Road Map by CMU Research Corporation’s Business Intelligence (BI) research committee work. The material covered integrates the committee’s efforts and strategies suggested in contemporary literature on how best to evolve an organization from a culture of purely intuition-oriented decision making to one that values and is capable of data-driven, fact-based, and model-assisted corporate decision making.

 

The Application of Mixture Designs for Formulation and Commercialization of Inkjet Inks

Bruce Knoebel
Senior Research Statistician, Eastman Kodak

Inkjet inks are comprised of a number of chemical components that are mixed together to produce a solution with certain physical, chemical, and image performance properties. Because the properties of an ink are determined by the relative proportions of the components in an ink, experimental designs and evaluations require the use of mixture design methodologies to maximize both their efficiency and their efficacy.

This talk will present the types of designs that have been used to identify an optimal ink formulation in research as well as the types of designs that have been used to demonstrate the robustness of an ink formulation for production.

While commercial software is used for the mixture design and initial analysis of the mixture data, custom in-house software has been developed for a more complete and in-depth analysis, understanding, and display of our data. The software uses a combination of Microsoft Excel (for data input and output tables) and SAS (DDE, Macro, IML, Graph) to generate the needed analyses and data displays. An overview of the numerical output and graphical displays from the software will be presented.

 

First Solar: Selecting Six Sigma Strategies that Shine

Steve Fowler
Director of Continuous Improvement, First Solar

When the business environment is dynamic and growth-oriented, not all Six Sigma tools can foster innovation and improvement. This session looks at what works – and what doesn’t – for First Solar, a maker of solar electric power modules. Examples include a practical robust optimization of a non-contact surface resistivity measurement system and the use of POV to prioritize further engineering characterization, root cause and design of experiments.

 

Modeling Economics 101: Innovation Risk Management

Michael Schrage
Researcher and Author of Serious Play: How the World’s Best Companies Simulate to Innovate, MIT

Building upon and extending the themes of the Monday morning keynote, this workshop goes into greater detail and depth as to how simple economic principles can help modelers get more value and more use from the models they build. Ideally, this session will be interactive and participants are expected to bring issues, concerns and examples for the group to discuss. Attendees should leave the session with next steps in mind and in plan.

 

Design of Experiments at Procter & Gamble

Kevin Norwood
Laundry Modeling & Simulation Research Fellow, Procter & Gamble

Cy Wegman
Corporate Modeling Simulation & Analysis Section Head, Procter & Gamble

Design of Experiments is an integral part of product and process development at P&G. DOE is used to build empirical models of product and process performance that drives better innovation, efficiency and speed to market. P&G uses a wide range of DOE, statistical analysis, visualization and optimization tools in JMP. This presentation will show some of the philosophy behind the use of DOE and give some examples of practical application.

 

Missing e-Bills Detection

Jane Damschroder
Senior Business Analyst, CheckFree Corp.

The Electronic Commerce division of CheckFree Corp. provides solutions that enable thousands of financial services providers and billers to offer the convenience of receiving and paying household bills online, via phone or in person through retail outlets. One of the solutions offered is the electronic delivery of bills via the Internet, a service currently provided at more than 2,000 sites. Through cooperative distribution agreements with over 450 billers, CheckFree delivers more then 58.7 million bills electronically each quarter. Replacement of paper bills with electronic delivery results in a significant cost savings to the biller and improved convenience and security to the consumer.

The goal of this project was to develop a system and processes that would allow proactive identification of potentially “missing bills.” Once a bill is identified as missing, the biller can send the affected bills to CheckFree and CheckFree, in turn, delivers the bill to the consumer, thereby avoiding any late payment issues.

The Missing e-Bills application is based on the Exponentially Weighted Moving Averages (EWMA) data model. The EWMA model is used to determine when unusual behavior is exhibited by a biller.

 

Maximizing Restaurant Capability: McDonald's Pursuit of Operational Excellence

Mike Cramer
Director of Operations Research for Worldwide Restaurant Innovation, McDonald’s

McDonald's is the world's largest QSR company with over 33,000 restaurants in 118 nations. Our quest for excellence as a service provider has propelled us to new innovative dimension, with a concentration on creating the "Flexible Operating Platform" that will enable strategic growth in all top markets. To reach our goals, we have invested in innovation and operations research. A significant part of that investment is the design and development of a unique portfolio of tools to accelerate operating platform design, development, testing and deployment. This portfolio includes Lean Six Sigma methodologies, video ethnography, data mining/analysis, engineered standards, dynamic ergonomic assessment and discrete event/agent-based modeling.

We will review our portfolio of tools and techniques and demonstrate how we have used these to make critical business decisions.

 

Building a Test-and-Learn Discipline at PNC

Sara Bennett, Eric Myers
PNC Financial

This paper will speak about our challenges, opportunities and successes in building a test-and-learn discipline within PNC. It will address such issues as cultural and organizational challenges, how we have approached these challenges and the results we are seeing. We will walk through several case studies to illustrate our journey and to highlight what has worked for us -- and what hasn’t worked. Finally, we will talk about what’s next in our world and how we hope to continue to develop analytical excellence within PNC.

 

Driving Business Strategy Through Rigorous Analytics

Chris Peterson
Senior Statistical Analysis Manager, Capital One Financial

Like many other industries, the financial services sector is highly competitive, heavily regulated and operates in a rapidly changing marketplace. However, the availability of vast data resources containing information about customer behavior implies that a significant competitive advantage can be achieved by effectively utilizing that information. To this end, Information Based Strategy (IBS) has been at the heart of Capital One's business philosophy since the early 1990's and an integral part of the company's success.

However, as the competitive landscape evolves the analytic competencies needed to sustain or grow any competitive advantage must also increase. This presentation will explore different ways companies can use improved data quality, data access, targeted testing, analytic tools and methods to drive business strategy and better compete in the future.

 

Using Monte Carlo Simulation to Predict Defect Rates for Process Control Scenarios in a Biological Process Using Online Process Analytical Technology

Byron Wingerd
Principal Scientist, Emergent BioSolutions

You may not be in the vaccine production market, but we bet you can benefit directly from lessons learned in how to improve production from pilot-scale to large-scale manufacturing. This talk will focus on issues with building models based on the variation in production data, validating the model predictions to actual data, employing the model with Monte Carlo simulation across design points that fill the space of all the Xs (space-filling designs) in your models and, finally, conducting what-if scenarios to determine process specifications for the scaled-up factory production process.

 

Follow What's Happening at the Innovators' Summit

Updates on the JMP Blog

Presentations, as available on the Agenda page

Summit Program

Download the Summit Program (3.69 MB, PDF)

Add your name to our interest list for 2009.

Thank you for your interest in the Innovators' Summit. If you'd like to receive e-mails about next year's event in Chicago, please provide your e-mail address below.

 

E-mail Address: