SeptOct2012 PDF

010101101010101010110101000100101010101 0101010 000100101010010010100100110101 010101010101011010101010101101010001001 010101010101010 0001001010100100101001 001101010101010101010110101010101011010 10001001010101010101010 00010010101001 001010010011010101010101010101101010101 RETAIL SCIENCE 0101101010001001010101010101010 0001001 DRIVING
PROFITABILITY WITH ANALYTICS 010100100101001001101010101010101010110 101010101011010100010010101010101010101 010001001010101010101010 0001001010100 100101001001101010101010101010110101010 10101101010001001010101010101010101000 ALSO INSIDE: 1001010101010101010 0001001010100100101 Analytics predicts presidential election Leveraging Big Data around customers 001001101010101010101010110101010101011 Certified Analytics Professional program 01010001001010101010101010101000100101
H T T P : / / W W W. A N A LY T I C S - M A G A Z I N E . O R G DRIVING BETTER BUSINESS DECISIONS S E P T E M B E R / O C TO B ER 2012
Executive Edge eBay VP Bob Page on the future of data and analytics
I N S I DE STO RY
Mudders uncover treasure

In the July/August issue of Analytics, John Toczek, author of the Thinking Analytically column in this publication and the PuzzlOR column for OR/MS Today, proposed a contest to, as he described it, put your analytical skills to the ultimate test. The contest, The 2012 Analytics Treasure Hunt, consisted of five puzzles that focused on common problems encountered in the mathematics, statistics and operations research fields. The answers to the puzzles, when properly solved and put together in the right order, revealed the coordinates for hidden treasure. To make things interesting, Toczek placed $100 of his own money in a small, metal tube at the secret location. Shortly after the issue went online, Toczek began hearing from hundreds of analysts/treasure-seekers from around the world interested in looking for the hundred bucks (hey, they do it for the ciphering, not the money). On Aug. 18 at about 10 a.m. EST, the team of Seneca Harberger (nom de blog Shepard Wallace) and Madineh Sarvestani retrieved the prize in Fairmont Park in Northwest Philadelphia; the coordinates: 40.07284,75.21907.
1 | A N A LY T I C S - M A G A Z I N E . O R G
The AIMMS PRO Platform A New Revolution In Optimization
Harberger, a student at Temple University Medical School in Philadelphia (which gave him a home-town advantage), and Sarvestani, who lives in State College, Pa., where shes a graduate student studying neural engineering at Penn State, have a common dominator: Harvey Mudd College in Claremont, Calif. Harberger holds a B.S. in physics from Harvey Mudd, while Sarvestani earned a B.S. in engineering at Harvey Mudd. Another Harvey Mudder, Tim Carnes, who completed his Ph.D. in applied operations research at Cornell and is now a research scientist at Link Analytics, introduced the two winners to the contest and helped verify their answers. You had a small crowd of people from Harvey Mudd looking for the treasure, Sarvestani tells us. To our fortune, most (of the Harvey Mudders) didnt live in the Philly area and none had my eagle vision. Toczek, who says his motive was to attract positive attention to the field of O.R. and analytics, plans to devise another contest next year.
Ever thought about ...creating an optimization App Store for business users? ...providing web access to your optimization solutions? ...deploying in the cloud?
No? We have! Check out our AIMMS PRO Platform at:
www.aimms.com/aimms-pro-platform
Norman Jerome, Aromatics Business Optimization Advisor - BP, plc. "In the first three months since migrating to the AIMMS PRO platform, we have seen our user base expand five-fold for one of our core scheduling applications. And with so many new eyes reviewing this application, the data are more reliable, and we are producing better schedules than ever before."
www.aimms.com/aimms-pro-bp
PETER HORNER, EDITOR peter.horner @ mail.informs.org
Bellevue, WA, USA Haarlem, the Netherlands Singapore Shanghai, China www.aimms.com info@aimms.com AIMMS is a registered trademark of Paragon Decision Technology B.V. Other brands and their products are trademarks of their respective holders
C O N T E N T S
XLMINER: VISUAL DATA MINING

Works with PowerPivot to Access 100 Million Rows in Excel
DRIVING BETTER BUSINESS DECISIONS
SEPTEMBER/OCTOBER 2 012
Brought to you by
FEATURES
9 DRIVING PROFITABILITY WITH RETAIL ANALYTICS By Sally Taylor-Shoff and Shalini Raghavan How to make choices and investments that deliver on expectations. 14 LEVERAGING BIG DATA AROUND CUSTOMERS By Rohit Tandon, Arnab Chakraborty and Ganga Ganapathi Opportunity to deliver real-time customer insights by harnessing the power of structured and unstructured data. 21 21 TAMING THE DATA TIDAL WAVE By Kevin Connor AHP combines business intelligence and data mining technologies with social aspects of human decision-making. 26 CERTIFIED ANALYTICS PROFESSIONAL By Scott Nestler, Jack Levis and Bill Klimack INFORMS prepares to launch rst-of-its-kind program. 30 PREDICTING THE PRESIDENTIAL ELECTION By Doug Samuelson Will quantitative historian and authors 13 keys once again open the door to the White House? 36 CORPORATE PROFILE: WALT DISNEY WORLD By Pete Buczkowski and Hai Chu How analytics enhance the guest experience at massive entertainment resort in Orlando.
Discover the easiest, fastest, lowest cost and risk on ramp to full-power data mining and predictive analytics: Analyze your data in Microsoft Excel with XLMiner software, now from Frontline Systems. You wont need a statistics PhD, months of learning time, or a huge budget for enterprise data mining software. But you will have the power of predictive analytics and data visualization at your fingertips. Sophisticated Data Mining Power. XLMiner goes far beyond other statistics and forecasting add-ins for Excel. It starts with multiple regression, exponential smoothing, and ARIMA models, but goes further with regression trees, k-nearest neighbors, and neural networks for prediction. It offers discriminant analysis, logistic regression, k-nearest neighbors, classification trees, nave Bayes and neural nets for classification, and association rules for affinity analysis. Practical Data Cleaning and Exploration. XLMiner includes utilities for data sampling and partitioning, missing data handling, binning, and
Tel 775 831 0300 Fax 775 831 0314 info@solver.com
transforming categorical data. You can apply principal component analysis, k-means clustering, and hierarchical clustering to simplify and cluster your data. Help, Textbook and Instructor Support. XLMiners extensive online Help introduces the data mining methods and explains the options in each dialog. With the popular textbook Data Mining for Business Intelligence built around XLMiner software you can go from beginner to expert business analyst in a highdemand field. Instructors can take advantage of lowcost classroom licenses, PowerPoint slides, example files, solutions to exercises, an online forum, and more. Download Your Free Trial Now. Visit www.solver.com/xlminer to learn more, register and download a free trial or email or call us today.
30
36
2 |
A N A LY T I C S - M A G A Z I N E . O R G
DRIVING BETTER BUSINESS DECISIONS
Brought to you by
RISK SOLVER PLATFORM
TEACH AND LEARN OPTIMIZATION AND SIMULATION
REGISTER FOR A FREE SUBSCRIPTION: http://analytics.informs.org INFORMS BOARD OF DIRECTORS President Terry P. Harrison, Penn State University President-Elect Anne G. Robinson, Verizon Wireless Past President Rina R. Schneur, Verizon Network & Technology Secretary Brian Denton, University of Michigan Treasurer Nicholas G. Hall, Ohio State University Vice President-Meetings William Bill Klimack, Chevron Vice President-Publications Linda Argote, Carnegie Mellon University Vice President Sections and Societies Barrett Thomas, University of Iowa Vice President Information Technology Bjarni Kristjansson, Maximal Software Vice President-Practice Activities Jack Levis, UPS Vice President-International Activities Jionghua Judy Jin, Univ. of Michigan Vice President-Membership and Professional Recognition Ozlem Ergun, Georgia Tech Vice President-Education Joel Sokol, Georgia Tech Vice President-Marketing, Communications and Outreach E. Andrew Andy Boyd, University of Houston Vice President-Chapters/Fora Olga Raskina, Con-way Freight
40
44
DEPARTMENTS
1 Inside Story 4 Executive Edge 6 Prot Center 7 Analyze This! 40 Conference Preview 42 Five-Minute Analyst 44 Thinking Analytically
Analytics (ISSN 1938-1697) is published six times a year by the Institute for Operations Research and the Management Sciences (INFORMS). For a free subscription, register at http://analytics.informs.org. Address other correspondence to the editor, Peter Horner, peter.horner@mail.informs.org. The opinions expressed in Analytics are those of the authors, and do not necessarily reflect the opinions of INFORMS, its officers, Lionheart Publishing Inc. or the editorial staff of Analytics. Analytics copyright 2012 by the Institute for Operations Research and the Management Sciences. All rights reserved.
INFORMS OFFICES www.informs.org Tel: 1-800-4INFORMS Executive Director Melissa Moore Meetings Director Teresa V. Cryan Marketing Director Gary Bennett Communications Director Barry List Headquarters INFORMS (Maryland) 7240 Parkway Drive, Suite 300 Hanover, MD 21076 USA Tel.: 443.757.3500 E-mail: informs@informs.org
The Excel Solvers Big Brother Makes Optimization Easier, More Powerful and Far More Visual. From its classic Solver Parameters dialog to its Ribbon, Task Pane, charts to visualize the shape of functions, Constraint Wizard, model diagnosis, Help, and coverage in leading management science textbooks, Risk Solver Platform makes it easier to learn and use optimization. Learn and Use Monte Carlo Simulation, Decision Analysis, and More with the Same Software. You can use its powerful optimization features alone but Risk Solver Platform is also a full-power tool for Monte Carlo simulation and decision trees, with a Distribution Wizard, 50 distributions, 30 statistics and risk measures, and a wide array of charts and graphs. Use the Full Spectrum of Optimization Methods. Fast linear, quadratic and mixed-integer programming is just the starting point in Risk Solver Platform. SOCP, nonlinear, non-smooth and global optimization are just
the next step. Easily incorporate uncertainty and solve with simulation optimization, stochastic programming, and robust optimization all at your fingertips. Go Beyond Single Optimization/Simulation Solutions. Define parameters in seconds, run high speed multi-way parameterized simulations and optimizations, capture all the solutions, and produce instant reports, charts and graphs, without programming. And you can do even more in Excel VBA. Free Trials and Support for Your Efforts at Solver.com. Download a free trial of Risk Solver Platform, and learn about the many other ways we can support your learning and teaching efforts all at www.solver.com.
ANALYTICS EDITORIAL AND ADVERTISING Lionheart Publishing Inc., 506 Roswell Street, Suite 220, Marietta, GA 30060 USA Tel.: 770.431.0867 Fax: 770.432.6969 President & Advertising Sales John Llewellyn john.llewellyn@mail.informs.org Tel.: 770.431.0867, ext.209 Editor Peter R. Horner peter.horner@mail.informs.org Tel.: 770.587.3172 Art Director Lindsay Sport lindsay.sport@mail.informs.org Tel.: 770.431.0867, ext.223 Advertising Sales Sharon Baker sharon.baker@mail.informs.org Tel.: 813.852.9942
EXECUTIVE EDGE
The future of data and analytics

Six futures of what eBay is working on today, all in various states of evolution.
From well-connected computers, social networks, mobile shoppers and the digitization of our lives to the ability to put sensors into and onto just about anything, we have a tidal wave of data. Mobile connectivity itself is driving a significant change in how we think about data. When a customer showrooms uses their phone in a retail store to compare prices online or uses their PC to purchase an item for after-work pickup at a local retailer the lines between online and offline start to blur. And therein lie both challenges and opportunities. As we saw in the July/August Executive Edge column from Piyanka Jain on predictive analytics, predicting the future relies on understanding the past. But there are times to break from it. Albert Einstein said, We cant solve problems by using the same kind of thinking we used when we created them, and taming the data deluge requires some creative thinking. For inspiration we turn to computer scientist Alan Kay, who said, The best way to predict the future is to invent it. eBays Marketplaces team (think eBay.com) has many tens of petabytes under management, with a constant stream of queries, tools and people that process and derive insights from the data. While job one is to keep up with the volume, variety and velocity (the 3 Vs) of incoming data, treading water doesnt make for a very competitive business. In order to provide as much
consumable intelligence to as many as possible, weve set upon a path to build and operate a data and analytics ecosystem that addresses these futures: 1. The future will happen in parallel. This isnt surprising, or even futuristic. Intel stopped making 80x86 processors in 2006, and even todays smartphones have multiple processing cores. Yet for too many years, and even in some organizations today, when you ran out of memory, storage, processing or I/O capacity, you bought a bigger machine and moved your workload. Several advances mean well start to scale out instead of always having to scale up. A small amount of capacity is cheap. Tie a lot of small boxes together and you have an enormous amount of capacity. The open source Hadoop environment makes it possible to harness that capacity in a coordinated way. Rather than move the data to the processor, we avoid moving data, which avoids the biggest expense. Instead systems like Hadoop move the processing to the data. With hundreds or even thousands of processors working simultaneously, were now able to do things we werent able to do before. 2. The future will be multi-platform. The Enterprise Data Warehouse (EDW) is no longer the canonical source of truth for everything, because were analyzing data that doesnt belong in a structured, relational system. We have semistructured data, like clickstreams, and data that appears to be unstructured, such as text, images or competitors Web pages. Different data
W W W. I N F O R M S . O R G
sets and different access patterns demand different systems. At eBay we have three major data systems: our EDW for transactional data, an enhanced SQL-like system for transactional + behavioral data, and Hadoop for everything. While this provides us with ultimate flexibility, its not without its costs not only for data replication but governance as well. Now metadata, data lineage and high-speed data transfer between the systems becomes much more important. Is it worth it? Absolutely, because today, no one system can handle the entire set of analytical needs. 3. The future will have machines that learn. Were seeing more machine learning lately. This is due in part to the sophistication of the toolkits that are emerging, but its also an acknowledgement that we have larger haystacks and smaller needles. We need machines to help sift through the data and help decide whats actually important. In a widely cited study, Microsoft researchers Michele Banko and Eric Brill (now head of eBay Research Labs) showed how five different natural language algorithms improved dramatically as they were fed more data. The lesson? In some cases, more data can trump new algorithms. That doesnt spell the end of the researcher, but it does show that improvements can come from multiple lines of inquiry. 4. The future will be self-serve. As data becomes integral to even the most mundane decisions, its important to allow the people with the questions to get at the answers themselves. While this has practical considerations
| S E P T E M B E R / O C T O B E R 2 012
BY BOB PAGE
A N A LY T I C S
THE MOST FLEXIBLE WAY TO DEPLOY YOUR MODEL

EXECUTIVE EDGE
SOLVER SDK PLATFORM
lower costs, faster time to insights it has an unexpected benefit: innovation. These new tools allow direct manipulation of data without having to translate between the person with he question and an analyst. Anyone at eBay can create his or her own analytics sandbox with one click. These virtual data marts allow rapid prototyping outside the main production environment, without requiring copies of data or rogue machines running under a desk. Another popular tool is Tableau Desktop, which makes interactive visualization within reach of many who would shy away from Microsoft Excel. 5. The future will be collaborative. Good analysis is a terrible thing to waste. What if you could surf the shoulder of a master analyst, or even your own analysis from a year ago, and leverage that for a current task? What if you could find others who had worked with similar data or business problems? At eBay we built what we call the DataHub a one-stop shop for all things data. It is a portal for publishing and finding data and analytics, but also has social-networking features such as friends/followers, groups and discussions. I can publish my Tableau or Microstrategy-based visualization or dashboard, and others can copy,
modify and comment on it. No more Excel/Powerpoint in e-mail! 6. The future will be live. Can you imagine trading stock solely by reading yesterdays closing price in the morning paper? Many business decisions are based on data even older than next-day, but organizations must be more agile. For a business like eBay, up-to-date means before the next click. Yes, some important data isnt updated in real-time, but the business shouldnt be held back by the slowest data feed. Yesterdays world ran on monthly or weekly sales reports, or even daily flash reports. But these analyses are a lens on the past. Running the business by examining the past is like driving a car by looking in the rear-view mirror. New tools such as memory-based data systems, Complex Event Processing and streaming query systems are getting mature enough to change business. So thats a quick half-dozen of the futures were working on today. All are in various states of evolution, but its clear that they will all play critical roles as we evolve the state of data and analytics. Its an exciting time.
Bob Page is vice president of Analytics Platform & Delivery at eBay.
The Optimization / Simulation Library that Speaks the Languages of Desktop and Web Computing. No other callable library matches the flexibility of Solver SDK Platform from solving Excel models on your server, to solving models created in your browser. Build Your Model Faster in Excel or Visual Studio. Use Risk Solver Platform or its subsets to build your model in Excel. Use Microsoft Visual Studio to create your model in C++, C# or VB.NET. Use C, R, MATLAB, Java, PHP or JavaScript its your choice. Use Optimization and Simulation as a Web Service. Run the Solver Server that comes with Solver SDK on any PC or server, then solve your models over the wire using our standard Web Service, SOAP and WSDL protocols. Get real-time progress, or just disconnect and retrieve the solution later its your choice.
Solve Your Most Challenging Linear Mixed-Integer, Nonlinear and Stochastic Optimization Models. Solver SDKs wide choice of built-in and plug-in Solver Engines handle every type and size of optimization problem, from Gurobi and Xpress-MP to Mosek and KNITRO. Use the SDKs built-in simulation optimization and stochastic programming algorithms, and support for multi-core CPUs. Download a Free Trial at www.solver.com. Learn more and download a FREE trial version of Risk Solver Platform, Solver SDK Platform, and our eight plug-in Solver Engines from Solver.com
PROFIT CENTER
Feeling the power
A perfect example of a management policy focused on blindly following prescribed metrics.
BY E. ANDREW BOYD
Management doesnt understand. So began the story of a power line technician I recently spoke with on a trip to California. Hed been at his job for decades and was one of the nicest people Id ever met. He wasnt expressing anger at management, just frustration that he wasnt able to do his job as efficiently and painlessly as he believed possible. His particular frustration grew from new metrics that had been put in place. His managers, I was told, were compensated based on how well the people who worked for them performed against these metrics. As a power line technician, he was measured by the amount of time he spent working in the field. Time in the office was considered wasted. He then shared a recurring problem brought about by the emphasis on time in the field. One of his duties was to disconnect power to homes and apartments when customers were delinquent in paying their bills. I went on to learn that this takes up a significant amount of his time, and, perhaps less surprisingly, it can be a dangerous activity. Flaring tempers and vicious dogs frequently factor into the equation. Over the years, hed learned it was worth the time to start his morning by personally
calling people he was scheduled to disconnect. It was thoughtful, but it served good business purposes as well. Customers wouldnt be surprised when he showed up, and he could gauge in advance how much trouble he might encounter when he stepped onto the property. But even more, faced with the realization that the power was about to be turned off (which entailed not only time without electricity but reconnection fees), many customers managed to arrange for payment before he showed up. The companys communications network was set up so that if payment was made even a few minutes before he arrived, hed know it and wouldnt turn off the power. His story seemed a perfect example of a management policy focused on blindly following prescribed metrics. Even if the phone calls led only a few people to clear up what they owed, the effort would pay for itself. Did the company think that showing up on the doorstep unannounced was good policy? Actually, no as I discovered from further discussion. Management had made an investment in an automated phone system that called delinquent customers at intervals leading up to the time of disconnection. Not all of the power line technicians were as diligent as my friend, and the automated phone calling was an effort to put best practices to work. Still, my friend wasnt convinced. People ignore automated calls he told me. When they speak with a live person, a person whos going to show up later in the day and turn off their power, it makes all the difference in the world.
He was convinced that his time on the phone was worth it. Who was right? I made a suggestion. If previous discussions with management hadnt achieved anything, why not show up with data? Spend one week making personal phone calls (on his own time if necessary) and another week not making them. For each week, tabulate the ratio of people who paid their bills before he showed up to the number of people scheduled to have their power turned off. If more people paid during the week he made phone calls, he had a case he could take to management. Its not certain hed get a positive response from those he worked for, but my experience is that when faced with numbers, rational people will sit up and take notice. When the argument is simply I know calling helps vs. I know calling doesnt help, its managements prerogative to do what it deems best. I was pleased when he took my suggestion seriously and, rather than push back, said hed think about giving it a try. I hope he does (and if I can track him down Ill share the results in a future Profit Center column). He may make a good case to management, or he may discover that the automated phone calls are doing a pretty good job. But either way, analytics will provide an answer. Thats the power of analytics at its best.
Andrew Boyd, senior INFORMS member and INFORMS VP of Marketing, Communications and Outreach, has been an executive and chief scientist at an analytics firm for many years. He can be reached at e.a.boyd@earthlink.net.
A N A LY T I C S
S E P T E M B E R / O C T O B E R 2 012
A N A LY Z E T H I S !
Colleagues, columns and catching up
There seemed to be a strong feeling in both industry and academia that an analytics revolution was indeed happening and I wanted to do my small part to document its progress and help both INFORMS and myself to find its place in it.
BY VIJAY MEHROTRA
Back in early 2010, I published the first edition of this column, a manifesto regarding what this series would be about. At that time, there seemed to be a strong feeling in both industry and academia that an analytics revolution was indeed happening and I wanted to do my small part to document its progress and help both INFORMS and myself to find its place in it. Since then, I have been looking for interesting stories in the world of analytics. This month, we check back in with some of these subjects to see whats happened since our last visit. When we last encountered my friend Hiram, he had been laid off from his job as director of analytics for a promising Internet advertising startup that seemed to have lost its way). Some of the symptoms: lack of communication between the algorithm developers and those working with customer implementations, obsession with producing test results to get (or keep) customers, spaghetti code as a result of years of on-the-fly development. Nearly two years later, I am pleased to report that Hiram has landed a job as a senior
director of analytics at another fast-growing platform company, and that he still seeks me out to serve as a sounding board from time to time. I am less pleased to report that he has started to see some of the same things most notably, skewed experiments designed to make the products performance appear better than it actually is at his new firm. Time will tell. Ben Alamar and Michael Lewis, the two most prominent people in my column on sports analytics, have both been on my mind lately. Alamar has been busy and productive; in addition to teaching his college students and writing a three-part series on the state of sports analytics [1] for this magazine with yours truly, Ben has also been quoted in the Wall Street Journal [2] and has been seen on the MLB Network describing his model for predicting the probability that a team makes the playoffs and/ or wins the World Series given the addition of a second Wild Card team in each league. In addition, he has been actively engaged in his work as director of Basketball Analytics and Research with the Oklahoma City Thunder, who played in the NBA finals this past June, and with working with ESPN to develop and test advanced football metrics. Interesting innovation on several different fronts. As for Lewis, I just finished reading The Big Short, his marvelous book on the world of subprime mortgages and the financial crisis that grew out it. From my perspective as an analytics professional, what really struck me about this book was how the quantitative
models were consistently used as fig leaves to rationalize faulty logic and ridiculously risky gambles and as deeply imperfect vehicles [3] for valuing financial instruments for which markets lacked any semblance of transparency. After visiting Ciscos Supply Chain Analytics group back in early 2011, I came away with a sense of the extraordinary scale of data management, modeling, analysis and organizational challenges that exist in a very large company that had done much of its growth through acquisitions. Since then, the company has restructured its entire supply chain organization, moved some of its analytics initiatives into functional areas such as marketing and operations and experienced some personnel changes all things that predictably happen in large companies. Most notably, Anne Robinson has not only become president-elect of INFORMS, but she has also left Cisco to join Verizon Wireless as director of Supply Chain Strategy and Analytics. A few months into her new role, I suspect that she is probably working her way through a whole new set of challenges given that shes joined another large and dynamic organization. On my summer vacation in 2011, I wrote a column about my visit to North Carolina States Master of Science in Analytics program. At the time, I had high hopes of starting a new interdisciplinary graduate program in analytics at the university where I teach. Alas, when we brought together folks from several different departments around campus to develop this new program, the process of starting this new
A N A LY T I C S
A N A LY Z E T H I S !
Advertising Opportunities at:
WWW.ANALYTICS-MAGAZINE.ORG
initiative turned into a bruising political battle. In the end, I found myself on the outside looking in. A profoundly disappointing experience but a valuable reminder about the challenges of innovating in collaboration with others. As the Jayhawks so succinctly put it, Its hard to sing with someone who wont sing with you [4]. Finally, last winter I had a chance to take a look at the good work being done by the Human Rights Data Analysis Group. Since then, others have also taken note of this organization and its unique capabilities. Most notably, HRDAG founder Dr. Patrick Ball was profiled in a recent article in Foreign Policy in which the author writes that his methods have changed our understanding of war. And by the time you read this, Ball and his colleague Megan Price will have conducted a large-scale sampling experiment to study the population found at Burning Man. (Through this project, they hope to learn how to gather accurate data in the context of a temporary community such as a sprawling refugee camp; since 60,000 people will be at Burning Man this year, this is a unique opportunity for this simulation. In addition, the HRDAG folks will help Burning Mans organizers conduct an analysis of the events demographics that takes into account non-response bias found in its sample data. Wait! Im supposed to be out there helping them with this experiment! I had better wrap it up for now. Perhaps Ill have more on this next time.
Vijay Mehrotra (vmehrotra@usfca.edu), senior INFORMS member, is an associate professor in the Department of Analytics and Technology at the University of San Franciscos School of Management. He is also an experienced analytics consultant and entrepreneur, an angel investor in several successful analytics companies and a longtime member of INFORMS.
Analytics: Driving Better Business Decisions provides a window into applications of mathematics, operations research, and statistics to drive business decisions.
REFERENCES
1. Part One: http://www.analytics-magazine. org/special-articles/391-beyond-moneyball-therapidly-evolving-world-of-sports-analytics-part-i Part Two: http://www.analytics-magazine.org/ november-december-2011/476-sports-analyticspart-2 Part Three: http://www.analytics-magazine.org/ march-april-2012/536-analytics-a-sports-part-iii 2. http://online.wsj.com/article/SB10001424052748 703421204576331173575453598.html 3. See p. 212-213 for a very memorable description of an argument between Morgan Stanley and Deutsche Bank in which each side cites its statistical model and underlying assumptions while arguing over a $1.2 billion trade settlement. 4. Blue, from the album Tomorrow the Green Grass (http://www.youtube.com/ watch?v=GUVVGQjWFg4)
Analytics
For more information about advertising, contact: John Llewellyn 770.431.0867, ext. 209 (888.3033.5639 Toll Free) Email: john.llewellyn@mail.informs.org
WWW.INFORMS.ORG WWW.ANALYTICS-MAGAZINE.ORG
Subscribe to Analytics
Its fast, its easy and its FREE! Just visit: http://analytics.informs.org/
CONSUMER INSIGHT
Driving protability with retail analytics

How to make choices and investments that deliver on expectations.
BY SALLY TAYLOR-SHOFF (LEFT) AND SHALINI RAGHAVAN

The most successful retailers today are increasing response rates to their offers and driving profitability by using Big Data and predictive analytics to make relevant, personalized, and precisely timed offers to customers. Predictive analytics provides a concrete means of realizing the long-standing exhortation to know your customer. In the era of Big Data, analytic tools have sufficient information to enable retailers to treat every
customer as an individual based on insights into their preferences and future behavior. Retailers who know their customers are making smarter decisions that maximize consumer loyalty without leaving money on the table by offering unnecessary or excessive rebates, discounts and special offers. These retailers are able to hit the sweet spot where customer behavior intersects with what the goals of retailers and their suppliers.
WHAT PROBLEM DOES ANALYTICS SOLVE FOR RETAILERS?
For retailers, one of the greatest values of analytics is providing decision points for determining how to treat each customer. For example, will it be profitable to offer Customer X free delivery? Or, would that offer be wasted because the customer is going to buy the product anyway? Consider what happens when consumers visit an online or brick-and-mortar store for the first time. The retailer knows
nothing about these potential customers and thus treats them all the same. At some point, the consumer may click on a product or category, or make a purchase. This consumer behavior is likely to trigger a business rule that initiates an action by the retailer. Coupons might be printed at checkout. An e-mail might be sent offering a discount on a product in an abandoned online shopping cart. In such instances, consumers are differentiating themselves by their behavior.
A N A LY T I C S
R E T A I L A N A LY T I C S
Because the retailer doesnt know anything more about these consumers beyond their click stream or the product they purchased, theres no reliable basis on which to make a more relevant offer.
The retailer, however, is still treating them all the same everyone who exhibits the same behavior receives the same offers. Inevitably the offers will be more relevant to some recipients than to others, and responses will vary accordingly. Because the retailer doesnt know anything more about these consumers beyond their click stream or the product they purchased, theres no reliable basis on which to make a more relevant offer. However, if analytically targeted offers can be made to consumers on a large scale, there is now a foundation on which to perform rapid test and learn. Both relevancy and speed to relevancy are critically important; they enable retailers to differentiate themselves in a crowded market to increase loyalty and margins.
UNDERSTANDING WHEN AND WHY TO USE VARIOUS ANALYTIC APPROACHES
and the printed coupons generated at instore checkout. The form of collaborative filtering most often used in retail is sometimes referred to as an affinity model or lookalike model. It identifies relationships between customers and items that have either been purchased or viewed, and infers how an individual will behave based on how other individuals who look similar (i.e., share one or more characteristic) have behaved.
Healthcare'13 Analytics half ad_Layout 1 9/3/12 5:49 PM Page 1
Collaborative filtering doesnt have to be triggered by a current transaction. It can be used to target ongoing campaigns and other kinds of promotions. Still, this analytic technique is fundamentally transaction-oriented. The algorithms used are best suited to modeling data about items purchased or viewed. Theyre not effective for modeling purchases with the wide range of bigger data retailers have in their databases (e.g., attitudinal data, seasonal purchase patterns, natural
INFORMS
INFORMS HEALTHCARE 2013

Join us in Chicago for the second INFORMS conference on healthcare O.R. and analytics, offering:
The best, most current research and applied work in healthcare O.R. and analytics in one highly-focused conference. Top-quality presentations by leading researchers and practitioners, selected through a review of extended abstracts. Cross-cultural view of healthcare systems and analysis of operational impacts. Structured networking opportunities, including birds-of-a-feather discussion groups and facilitated networking over lunch. Intimate, small-scale setting in a convenient and dynamic city.
2ND CONFERENCE ON HEALTHCARE Spring 2013 (dates to be announced) Chicago, IL
Lets look at what different types of analytics can tell retailers about their customers, giving them various decision points to consider when determining which actions to take. Collaborative filtering enables retailers to take a degree of targeted action even with first-time customers. This type of analytics is often behind the product recommendations offered on e-commerce sites
Help Promote Analytics

Its fast and its easy! Visit: http://analytics.informs.org/button.html
www.informs.org
Watch for more information and conference dates.
10
A N A LY T I C S
product adjacencies, basket builders) or can access from external sources (e.g., demographics, public records, third-party marketing information). Clustering algorithms enable retailers to differentiate between customers in broad ways such as customers who like premium brands or customers who prefer organic food. One of the benefits of painting customers with this kind of broad brush is that it can help direct and justify large-scale expenditures on store design, new merchandising schemes and promotional programs. This approach to analytics moves closer to the use of Big Data. In fact, this approach can utilize exceptionally large data sets and a wide range of data types. Using analytics in this way (often called behavioral segmentation) en ables retailers to make far more accurate decisions than can be achieved through traditional methods of database querying on customer attributes such as recency, frequency and monetary value of past purchases. Ana lytics is more accurate in large part because it can handle greater data complexity. While query-based seg mentation generally involves no more than three to six customer attributes,
analytic-based segmentation can encompass dozens or even hundreds of attributes, greatly expanding the range of possible segmentation schemes down to a nearly individual level. Most clustering methodologies try to make sense of these hundreds of attributes analytically. A select few clustering algorithms use a hybrid approach of balancing the analytic approach with inputs from the business user. This ensures that the algorithmic approach is one that is more usable by the retailer. With many more ways to group customers, and the ability to try lots of alternative groupings quickly, retailers can make better strategic and resource-allocation decisions. One large national retailer, for example, has used analytics-driven segmentation to better understand and serve customers who account for the bulk of the companys revenues. Using their characteristics to define population segments, and using these segments to guide decisions from store layouts to how staff interacts with customers, this retailer increased same-store sales in the first quarter of implementation alone by 8.4 percent resulting in a 15 percent increase in total revenue.
269
What would you do with an extra 269 minutes?
A leading marketing companys scoring models used to take 4.5 hours to process. Now, with high-performance analytics from SAS, theyre scored in 60 seconds.
Get the relevant insights you need to make decisions in an ever-shrinking window of opportunity and capitalize on the complexity of big data to differentiate and innovate.
high-performance real analytics A game changer.

High-Performance Computing Grid Computing In-Database Analytics In-Memory Analytics Big Data
sas.com/269 for a free white paper
Each SAS customers experience is unique. Actual results vary depending on the customers individual conditions. SAS does not guarantee results, and nothing herein should be construed as constituting an additional warranty. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. 2012 SAS Institute Inc. All rights reserved. S91566US.0412
Large retailers who serve tens of millions of customers, each of whom has many attributes and preferences, can go as far as to essentially create segments of one.
Join the Analytics Section of INFORMS

For more information, visit: http://www.informs.org/Community/Analytics/Membership
Propensity models enable retailers to predict how individual customers are likely to behave. With such spe cific insights, retailers can differentiate between customers to a much greater degree, further increasing the granularity of segmentation and the relevance of offers. This analytic approach is particularly effective in Big Data environ ments. In fact, large retailers who serve tens of millions of customers, each of whom has many attributes and preferences, can go as far as to essentially create segments of one. Propensity models deliver this level of specificity and accuracy because of the ability to handle enormous amounts of internal and external data. Further, they are able to pinpoint the specific customer attributes in the data that are most predictive of a future behavior. Relationships between numerous attributes and other variables are examined to see how a change in the value of one variable affects the value of another. Attributes that prove to be highly predictive of a behavioral outcome are incorporated into a predictive model. For example, a propensity model can be built to predict a customers propensity to make a specific purchase or to discontinue using a premium service. Because these models can make
predictions for individual customers, they open up the possibility of unique treatment. Moreover, by using multiple propensity models, retailers can gain a much clearer picture of the customer. Knowing that Jane is not only likely to buy a 48-inch TV, but that she tends to like the Sony brand, but she doesnt tend to pay premium prices for cutting-edge technology, enables the retailer to greatly increase the relevance of individualized offers and interactions. Here are examples of how leading retailers are applying such insights: A large retailer is using this type of analytic approach to increase the ROI from promotional campaigns. When a popular new movie or video game comes out, for example, the retailer sends offers only to those likely to buy the product within the offer redemption period. Response rates are two to three times higher than when the same offer is sent to everyone. And because the retailer is not wasting customer time with irrelevant offers, future promotions are likely to be considered by the customer. Another large retailer is improving its ability to predict when customers are about to make a big purchase by incorporating customer clickstream data into propensity models. Many
consumers do extensive online research before making a major purchase. By analyzing billions of clicks across millions of customers, along with each customers purchase histories and historical behavior patterns, the retailer can pinpoint the right moment to make an offer.
UPLIFT MODELS
If a propensity model predicts when a customer is likely to buy a given prod uct, why should the retailer go to the ex pense of sending a promotional offer? Uplift models help retailers determine if an investment is likely to be worth the result. Often used in conjunction with propensity models, uplift models predict the amount of change likely to occur in customer behavior as a direct result of a particular retailer action. Uplift models can save retailers millions by enabling them to avoid offering discounts to customers who will purchase without them. For example, such a model might predict whether or not a 20 percent discount is likely to increase a particular customers propensity to buy a pair of designer jeans within the next two weeks. The retailer can then send the coupon only to customers whose behavior its likely to change.
12
A N A LY T I C S
When uplift modeling indicates a customers behavior is likely to be affected by a promotion, it can also help retailers determine which promotion will have the most impact. Will 20 percent off be any more effective than 10 percent off? Will any discount be more effective than free shipping? Is offering 12 months of interest-free credit necessary, or will six months be nearly as enticing? Uplift models provide the analytic insights retailers need to make precise decisions about where to put marketing spend for higher ROI. Uplift models are based on Big Data analytic techniques that can predict individual customer sensitivi ties to price incentives, redemption terms and even promotional package design. For instance, one company that helps to make markets for new products by spending heavily on pro motion, is using uplift models. The analytic insights enable the retailer to accelerate the purchasing behavior of so-called laggerscustomers who historically havent been among the first to purchase. By targeting these customers with offers that are likely to change their historical behavior, the retailer is increasing the concentration of sales in the first two months of the product lifecycle its critical
period before competitors can draft off of their momentum.

CONCLUSION
How will you stand out from the crowd?
As retailers add more data to their analytic efforts, they increase the number of decision points for differentiating between customers and making more targeted decisions. But just as having lots of data can be overwhelming and of little value in and of itself, so it is with the analytic predictions drawn from Big Data. Their business value depends on the retailers ability to operationalize them. Retailers of all sizes are bringing analytics into their operations. The key to making choices and investments that deliver on expectations is to understand what various types of analytics do and how they fit (or dont fit) what the business is trying to accomplish. Its also important to think about analytics as an incremental process rather than a packaged solution. No one approach serves all purposes. Wherever a company is in the spectrum of analytic sophistication and experience, theres a next step to take to achieve even greater benefits.
Sally Taylor-Shoff is a vice president in FICOs Marketing Practice. Shalini Raghavan is director, analytic product management, at FICO. Together they have nearly 40 years of experience in retail and marketing analytics.
A membership in INFORMS will help!
New in 2013! Certification for Analytics Professionals Online access to the latest in operations research and advanced analytics techniques Subscriptions to online and print INFORMS Publications Networking Opportunities available at INFORMS Meetings and Communities Education programs around the world to enhance your professional development and growth!
begin online September 11
INFORMS Renewals
2013
Renewing online is quick and easy. Renew online before October 5, 2012, and be entered to win an iPad.
visit http://join.informs.org after September 11 to join online
M A R K E T I N G : DATA I S D I V I N E
Leveraging Big Data around customers

Opportunity to deliver real-time customer insights by harnessing the power of structured and unstructured data.
BY ROHIT TANDON, ARNAB CHAKRABORTY AND GANGA GANAPATHI (LEFT TO RIGHT) Consumers are exercising freedom of choice regarding numerous online services, making them far more empowered.
The rise of the Internet has created terabytes of data or Big Data that is available to consumers and enterprises alike. This digitization is an emerging phenomenon and is pushing the boundaries of every industry, shifting consumer buying patterns. Marketers are leaving no stone unturned the products that could once be picked up off the rack are now sold online. To their advantage, the consumer is exercising freedom of choice regarding
numerous such online services, making consumers far more empowered. Consumers research a prospective purchase offline as well as online, migrate between channels in the course of a single buying cycle, voice unsolicited opinions safe in the knowledge that these may be heard by organizations and influence buying behavior of others in their network, reiterating that word-of-mouth has gone digital. The sophistication of buying behavior is at unprecedented levels, and it will only increase going forward.
Some concepts to keep in mind: Data is divine: About 80 percent of data is unstructured. Not only does this data open new avenues of analysis, thus allowing companies to come closer to its customers in real time, it can also lead to premature death of companies if they are not receptive to the data to understand customer needs. By successfully understanding data, companies will be able to identify trends, create new products and services, and in the long run,
14
A N A LY T I C S
L E V E R AG I N G B I G DATA
make decisions and strategies that are backed by pure data and not sheer intuition. Equipped with large amounts of customer data insurance companies, retailers, transportation companies and communications providers have a unique opportunity to make this type of information services play to their advantage. Sentiment-al: While companies dont dispute the importance of data, they face maximum challenges in extracting and comprehending the data. Probing the unconscious mind of the consumer has tremendous value beyond advertising. It reveals that what consumers actually believe or think may contradict what they say when asked. Global e-commerce sales may soon reach $1 trillion. This trend is directly related to the number of people making transactions online, creating an opportunity for companies to use the online information to assess customers buying behavior and clout to make or mar a brand. Marketers can add immense
value by making observations, listening and acting rather than focusing on invasive advertising. This will be the game changer. Imagine, at a simplistic level the launch of a campaign for a new product or service. Paid advertising per forms its job of awakening interest and creating desire for a hitherto unarticulated or unrecognized need. This might lead to a couple of searches, posts on social media platforms, and, if lucky, a visit to the company or brands Web site. The propensity to purchase is probably increasing or decreasing with every new byte of information consumed, and all this happens without even one tradition ally defined formal interaction with the company. The evangelists or de tractors impacting this decision are individuals who want to share their experience as a shopper and help a fellow shopper. Imagine the potential of growth or decline a single customers experience and comment online can create for a company.
NORTHWESTERN ANALYTICS
As businesses seek to maximize the value of vast new streams of available data, Northwestern University offers two masters degree programs in analytics that prepare students to meet the growing demand for data-driven leadership and problem solving. Graduates develop a robust technical foundation to guide data-driven decision making and innovation, as well as the strategic, communication and management skills that position them for leadership roles in a wide range of industries and disciplines.
MASTER OF SCIENCE IN ANALYTICS 15-month, full-time, on-campus program Integrates data science, information technology and business applications into three areas of data analysis: predictive (forecasting), descriptive (business intelligence and data mining) and prescriptive (optimization and simulation) Offered by the McCormick School of Engineering and Applied Science www.analytics.northwestern.edu MASTER OF SCIENCE IN PREDICTIVE ANALYTICS Online, part-time program Builds expertise in advanced analytics, data mining, database management, financial analysis, predictive modeling, quantitative reasoning, and web analytics, as well as advanced communication and leadership Offered by Northwestern University School of Continuing Studies 877-664-3347 | www.predictive-analytics.northwestern.edu/info
Request a no-obligation INFORMS Member Benefits Packet

For more information, visit: http://www.informs.org/Membership
15
appropriately will reveal insights that can separate the marketing seers from the mere also-rans. Seeing this deluge of data as a fountainhead of insights will help understand the present and foretell the future in a way that can creFigure 1: Big Data is within grasp of organizations that can tame the ate competitive advandata. tage like never before. For example, it can Big Data is within our grasp: Therefore, improve access to articulated and unarall marketers require is a vision to tame ticulated customer needs, make reveladata that millions of such interactions tions of need gaps, identify influencers, and transactions create. One also has to tell the story behind trends, confirm early keep in mind the existing structured data, warning signals; in short, the power that which together with unstructured data marketers in turn can wield to create cuscan create a new meaning to marketing. tomer advantage has never been this obCompanies that realize this early will be vious before. true pioneers of the Big Data era. Data points from multiple sources viewership Customer insights for ROI (Its all data, search, clickstream, social media about the money): With most transacmetrics & sentiment, e-commerce, offline tion data, and even some forms of unretail if stitched together and analyzed structured data being largely understood today, insights that will emerge from interactions that havent been documented in as much detail yet are next in line, such Subscribe to Analytics as shop floor data to discern insights from Its fast, its easy and its FREE! sales-rep interactions, emotion analysis Just visit: http://analytics.informs.org/ during all human interactions, and other
voice and video data for possibly unspoken revelations. These, when combined with better understood forms of data, will begin to reveal insights that can be applied across the marketing value chain. Going forward, decisions on product roadmaps, marketing messages, sales levers and service interactions will become so steeped in insights that hitting the bulls eye will cease to be matter of chance. Insights will have a direct impact on ROI. For example:
by uncovering new markets demographic or psychographic customer segments for new and existing propositions; by helping make sales cycles shorter and more effective, thus optimizing investments; and by making service interactions effective, efficient and delightful for the customer, reducing dissonance and increasing loyalty. Driving the CI function to its rightful place under the sun.
LOOKING? HIRING?
http://jps.informs.org For many years, INFORMS has offered a Job Placement Service to connect employers and qualified O.R. and analytics professionals. This service can be used alone or in conjunction with the Annual Meeting Job Fair at the 2012 Annual Meeting. Both give applicants and employers a convenient venue for meeting. JPS is free to INFORMS member applicants. For info on job fair at the 2012 INFORMS Annual Meeting, visit http://meetings.informs.org/phoenix2012/jps.html
A N A LY T I C S
Customers want to be recognized for the individuals they are and not treated as one among a hundred others.
Figure 2: Listening to the customer.

Instances abound for the way Big Data can be applied across the marketing value chain. The utility of social media analytics spanning the functions of insight generation, engagement, health metrics when combined with campaign information, macro-economic data, competitive data, sales insights, customer service interactions and much more on an on-going basis, can yield real-time, actionable insights that were previously difficult to sustain on an on going basis. For example:
Product teams can get rich insights to plan with: Freeing themselves from the constraints of OEM blueprints and short-term demand plans, product teams can have access to the exciting insights theyve always dreamed of basing their innovations on. While primary research and observations may continue to be the mainstay for some time to come, the scalability of listening in to social media platforms has brought the world closer than ever before. Emerging markets will never open up using the previous years trends from developed markets, and exciting idea for new categories itself may emerge from robust listening programs. Marketing is as much about the relationship as it is about the transaction: Customers want to be recognized for the individuals they are and not treated as one among a hundred others. Having a Web page that knows that
customers have returned for a second look, understands that desire is driven by an intimate understanding of the customer psyche and offers a replacement just when customers are contemplating a switch these are the hotspots that turn customers on and buys their loyalty. In addition, the brand evangelists will become the trusted third-party influencer to others contemplating a relationship with the offering. The most compelling argument, of course, is the opportunity to optimize marketing spends based on reading social signals correctly. Returns from bought and owned media begin to plateau or even show negative returns beyond a point, but juxtaposing this impact with
Figure 3: The insights data can provide are unlimited.

A N A LY T I C S | S E P T E M B E R / O C T O B E R 2 012
17
the positive mileage that earned media can deliver recognizes the opportunity to save precious marketing dollars without compromising impact. Spray & pray replaced by sure shot targeting: In this era of limited everything investments, sales teams, time the only things that need not be limited are the insights that data can provide. Combining hitherto disparate data sets and analyzing them in ways that were previously not possible to drive insights around sales personalization is possibly the lowest hanging fruit in an enterprise. The work is with largely recognized data sources, the returns are almost immediate, and the ROI delivered eases the next battle that will be fought to justify Big Data investments. Real time insights for better customer service: Technology has enabled customer service at the touch of a button and is no longer dependant on the last mile interaction at a service center. Service interactions fit into the Big Data world in two wonderful ways: using insights generated from correctly assembled and analyzed Big Data, interactions can be personalized with a deep understanding of the customer and with a high chance of guaranteeing satisfaction. For instance, concerns
that begin to trend in the social world soon enter call center interactions. Prepping agents or using the early warning signals to take decisive action are signs of an enterprise reacting in real time. On the flip side, all interactions, whether on chat or voice, are rich fodder for enriching the database. For example, call centers can analyze text, sentiment and emotion combined with trends, demographics and psychographics to paint multi-dimensional pictures of the target audience, the competition and market behavior.
REAL-TIME DEPLOYMENT OF CUSTOMER INSIGHTS
WHATS AN ANALYTICS MAGAZINE READER TO DO AT THE 2012 INFORMS ANNUAL MEETING?
ATTEND AN ANALYTICS SESSION

Analytics for a Networked World Edelman Finalist Presentations The Role of Decision Professionals in the Analytics Movement Predictive Analytics Applications Developing a Healthcare Cost Index Using Advanced Analytics Tutorial: Analytics in Consumer Credit Challenges for Public Sector Analytics Business Analytics Curriculum in Higher Education Predictive Analytics for Solid Organ Transplant Operations Big Data Analytics for Condition Based Maintenance Business Case studies in the Petrochemical and Petroleum industries Innovation Analytics: Using Complexity Science and Big Data for Sequencing of Products Service Delivery Modeling and Optimization Managing Immigration and Customs Enforcement's Program Operations with Innovative Analytics Robust Routing for Battery Electric Vehicles BracketOdds: Examining the NCAA Mens Basketball Tournament with Advanced Analytics Machine Learning for Power Grid Reliability: Predicting Manhole Events in New York Analytics for OR Access at a Large Teaching Hospital Roundtable on Analytics and Intermodal
MEET ANALYTICS SECTION MEMBERS

The Analytics Section Business Meeting will be held Monday, October 15, 6:157:15p.m. in Phoenix Convention Center.
The definition of Big Data is bound to vary between organizations and the quantum and variety of data that they have been used to in the past. For any enterprise CTO, the first implication of Big Data is the need for a strategy for dealing with large quantities of data. The term Big Data encompasses complex data sets so large and unwieldy that existing database management tools cannot aid in the collection, storage, analysis, sharing and visualization of the raw data and insights generated. The large data sets get generated because Big Data is a supposedly large, related data set, as opposed to the
RECEIVE INDIVIDUAL CONSULTING

Analytics Magazine readers are invited to meet one-on-one or in small groups with Analytics Section officers in Phoenix. Well help you arrange. Email gary.bennett@informs.org for details.
REGISTER HERE
meetings.informs.org/phoenix2012/register.html
Figure 4: The technology stack.
smaller component data sets that exist in isolation. This allows for co-relations to be found between data sets that could not previously talk to each other. The technology stack has to be such that it allows for the collection and storage of data along the three dimensions of challenges and opportunities that it presents: the increasing and ever-growing volume of data, the velocity of data (in and out) and the variety across the types and sources of data. Customer-related data for an enterprise may come from many sources:
Web logs, social data, Internet search indexing, call center call and chat records, video archives, e-commerce, sales transactions, macro-economic data, related category market, competitive data, etc. By present standards, the volume of business data worldwide across all companies doubles every 1.2 years, so the CTO needs to have a vision that takes into account the growth of data that the enterprise will need to be prepared for. The key features that typically form the basis of a Big Data analytics platform may include; real-time query and loading enables immediate access for rich analytics;
in-database analytics eliminates the need to extract; database designer tools with the ability to work with structured and unstructured data, and enabling continual improvements while the system remains online; data compression and columnar storage and execution for faster and efficient querying; ability to scale out on demand without adding expensive infrastructure; high availability; optimized workload management so the users focus on execution; and seamless integration with the growing ecosystem of analytics solutions such as Hadoop, MapReduce, ETL and Native BI. Deployment on the cloud is also gaining acceptance, and this may help smaller companies with limitations on scalable infrastructure. The techniques and technologies that need to be enabled to actually process Big Data are a combination of traditional and new-age methods: A/B testing, association rule learning, classification, cluster analysis, natural language processing, neural networks, pattern recognition, predictive modeling, regression, sentiment analysis, time series, etc.
The last mile, and often the most critical one, is possibly visualization. This is a challenge for Big Data analytics, and were seeing continued evolution in this space. Better visualization tools are critical because many cases of Big Data analytics deployment demand instant time to reaction, making the case for increased efficiency. Also, those who finally execute based on the insights may not necessarily be data scientists, and this makes it all the more important for findings and recommendations to be presented as clearly as possible. Needless to say, Web and mobile platforms are equally important to the final consumption of analytics.
THE FUTURE OF BIG DATA
The future of customer insights has never been brighter, and it will only scale greater heights from where it is right now. In this era of limited everything investments, sales teams and time the only things that need not be limited are the insights that data can provide. Combining hitherto disparate data sets and analyzing them in ways that were previously not possible will drive insights around sales personalization, and it represents possibly the lowest hanging fruit in an enterprise. The returns are almost immediate and the ROI delivered eases the next battle that will be fought to justify Big Data investments.
A N A LY T I C S
L E V E R AG I N G B I G DATA In this era of limited everything investments, sales teams and time the only things that need not be limited are the insights that data can provide.
While primary research and observations may continue to be the mainstay for some time to come, the scalability of listening in to social media platforms has brought the world closer than ever before. The utility of social media analytics spanning the functions of insight generation, engagement and health metrics, when combined with campaign information, macro-economic data, competitive data, sales insights, customer service interactions and much more on an on-going basis, can yield real-time, actionable insights that were previously difficult to sustain on an ongoing basis. The digitization of life that we see around us provides millions of data points that can be stitched together to provide 360-degree views of the customer. Thinking ahead and being prepared to satisfy unarticulated needs through real-time insights and customized recommendations will help create customer delight like never before, as well as competitive advantage that can be quickly monetized. The overall story is a win-win situation for the both sides of the table, leaving us with the obvious answer to one last question: Has there ever been a better time to be a marketer?
Rohit Tandon, senior INFORMS member, is the vice president and worldwide head of Global Analytics at Hewlett-Packard Company, where he helps drive the analytics ecosystem to support HPs vision and priorities through delivery of cutting edge analytical capabilities across sales, marketing, supply chain, finance and HR domains. Arnab Chakraborty (arnab.chakraborty@hp.com), senior INFORMS member, is director of Global Analytics at H-P, where he is responsible for partnering with senior executive leadership teams within HP and drives the deployment of analytics solutions across the Americas, EMEA and APJ regions. Ganga Ganapathi (ganga.ganapathi@hp.com) is manager of Global Analytics at HP, focusing on marketing analytics. She partners with business teams to drive deployment of analytics solutions across the Americas, EMEA and APJ regions.
Decision Sciences Institute

advancing the science and practice of decision making
The Decision Sciences Institute is a nonprot professional organization of researchers, managers, educators, and students interested in decisionmaking techniques and processes in private and public organizations. The Institute is an international organization with over 3,500 members in 32 countries. The annual meetings and regional conferences attract over 3,500 participants a year. The Decision Sciences Institute Is Committed to . . . Research. A focus on the integration of research in the art and science of managerial decision making across traditional functional academic disciplines; an international forum for presentation and discussion of research. Teaching. A forum for presentation and discussion of innovative teaching; recognition of teaching excellence and curriculum innovation. Practice. An exchange of ideas between leading professional practitioners and educators.
Benets Members Receive: Annual meeting. Members receive discounted fees to attend, and the meeting draws over 1,500 professionals together to share current thoughts on theoretical and applied issues. Decision Sciences is a highly respected journal among scholars and is subscribed to by over 1,000 libraries. It seeks and publishes high quality, theoretical and empirical articles. Decision Sciences Journal of Innovative Education is a high quality, peer-reviewed scholarly journal whose mission is to publish signicant research. Decision Line, the Institutes news publication, is published ve times annually and includes feature columns, as well as information on members, regions, annual meetings, and placement activities. Job Placement Services are offered throughout the year, with position and applicant listings available via the Internet. Also at the conference, special facilities aid in position search activities.
Join us.
Log on to our website and complete the membership application to begin enjoying your benets. Regular membership: Student membership: $160/year $25/year
n n n n n n n n n n n n n n n n
Decision Sciences Institute

43rd ANNUAL MEETING
NOVEMBER 17-20, 2012
n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n
www.DecisionSciences.org n n n n n
n n n n n n n n n n n n n n n
20
DECISION-MAKING
Taming the data tidal wave

AHP combines business intelligence and data mining technologies with social aspects of human decision-making.
BY KEVIN CONNOR
Nothing changes the course of world history, public policy and business results like decision-making. Good decisions are based on an in-depth understanding of the benefits, opportunities, costs and risks inherent in our choices. To make the best choices, decision-makers need information. Those decisions are hampered by the growing tidal wave of undifferentiated information that is the sign of our times.
By some estimates, the amount of digital information is increasing tenfold every five years. We have reached an amazing point in human history: We are collecting so much information that it is impossible to know what we know. Hidden among the bits and bytes, encoded as zeros and ones, are trends, connections and insights that both define and shape our world. How can businesses and government institutions find meaningful and actionable
21
A N A LY T I C S
A N A LY T I C H I E R A R C H Y P R O C E S S
insights to inform decisions and improve performance? This article looks at one way to uncover those trends, connections and insights: a mathematical principle called the Analytic Hierarchy Process (AHP). Developed in the 1970s, AHP is now being integrated into complicated decisionmaking processes by both public sector and private industry.
THE INFORMATION TIDAL WAVE: SCOPE CREATES UNCERTAINTY
Some recent studies estimate that the average knowledge worker now spends two hours a day searching for information later deemed useless, while an estimated 80 percent of this wealth of digital information is unstructured. Couple these ideas with studies that suggest one-third of our decisions arent implemented and that half of those that are dont meet their objectives.
Were capturing more information than ever before, but were not really using it to
our advantage. All of this leads to choices that seem more uncertain, increasingly complex and risky. With all this information at our disposal, we feel we should know, yet we dont. Group decision-making is difficult. Advances in technology to improve our use of data are evolving rapidly, but integration of data with the human and social elements of decision-making have continued to prove challenging. How can organizations bring together people, establish processes with the appropriate tools and technologies to facilitate efficient and effective decision-making in the world of voluminous data all while honoring the intuition and judgment of their best minds? Dealing with the cresting wave of information requires that we take a broader view of decision-making that brings together both the best information, subject to the best minds, through a process that integrates them. By blending the most relevant information with the wisdom and intuition of experts, our ability to tip the scales of uncertainty and risks toward benefit and opportunity can be greatly increased. So, how might we find such a process that allows us to utilize the appropriate content we have and give context from our expertise and knowledge? Some of
the best minds in the public sector and private industry depend on the AHP for surprising insights.
BEFORE THE ANALYTIC HIERARCHY PROCESS: THE VOLATILITY OF GROUP DECISIONS
Dr. Thomas Saaty developed the AHP at the Wharton School at the University of Pennsylvania in the 1970s, while working on nuclear non-proliferation negotiation strategies for the State Department. In aligning the perspectives of some of the worlds brightest economists, utility theorists, game theorists, scientists and lawyers, Dr. Saaty came to a realization: Although this group of brilliant people had tremendous intellectual gifts and knowledge, their challenge finding a coherent and aligned point of view was a social one. The fear of not being heard or understanding how our contribution influences a choice can bring out the worst sides of our nature when interacting collaboratively to choose a course of action. Battling it out in decision-making often reduces collective discourse in organizations and societies from a fair, rational, respectful and transparent debate about what is best for us to an advocacy based, power driven, contemptuous and in the worst cases dishonest grab for what is best for me.
Whether we are buying a home or a car with our spouse and family, attempting to understand the most probable risks to national security, or choosing the next blockbuster drug to improve the health and well being of society, subtle and potentially corrosive threads can infect any decision. Our inability to see what others see and why, or that others perspectives may be valid, tends to more deeply entrench our own biases, and forces us to vehemently defend our position. Along the way, this approach destroys value and compromises our ability to achieve our goals. How can any group, with its disparate and diverse views of the world, express what was important to each member in addressing the challenge at hand, and be able to see their collective and desired course of action? AHP is a collaborative decision-making methodology that is used to structure and analyze complex and potentially volatile decisions. Based in fundamental principles of mathematics and psychology, it was developed for the purpose of synthesizing the judgments of a group of decision-makers and providing them a rational, transparent and collaborative way to express themselves and understand the totality of their interests, preferences and priorities.
A N A LY T I C S
THE INNER WORKINGS OF AHP: MERGING DATA AND EXPERT JUDGMENT
The application of AHP follows a structured format. Lets outline the key steps in the process: Define relevant criteria: Decision-makers develop a hierarchy or tree of the relevant criteria, objectives or goals for their decision. The criteria are grouped in clusters from high-level categories at the top level to more specific sub-criteria that define those categories at the bottom.
These lowest level criteria are then described with ratings and measures that will be used to differentiate the value of options against these criteria. This step creates the scorecard that captures data and evaluations. These criteria can range from the purely quantitative factors such as net present value or miles per hour, to more subjective and qualitative factors such as strategic fit or risk. Establish the weight of each criterion: It is important to establish the relative priorities of the criteria to be used for assessing the options. Having been identified by the group of decision-makers, the criteria are all at least relevant and of interest but clearly not equally important. A group of decision-makers must determine which factors more strongly predict a valuable outcome in their decision given the context of their choices. By comparing each the criteria against each other using a pairwise comparison approach, decisionmakers are forced to value the decision criteria head-to-head to deal only with the difference in each paired combination: Which is more important? By how much? Why? A or B?, B or C? And so on.
Figure 2: Sensitivity analysis clarifies decision alternatives.
Mathematics can then determine how all the criteria relate amongst each other. By stating their preference openly, decision-makers can identify points of agreement and disagreement and have a framework for expressing information that only they may have that is influencing their position on some matter. Evaluate options based on these ratings: Decision-makers (or the subject matter experts that they entrust) must rate or evaluate the options using the quantitative or qualitative rating scales that were derived to measure how well the options reflect the priorities expressed in the criteria. This
results in a relative prioritization or score for the various options that now includes the blended priorities of the decision-makers and reflects the impact of the known measures or metrics as well as the judgment of the experts that are party to the decision. Consider alternatives: The group may perform sensitivity analysis on the prioritization of the decision alternatives for example, what if criterion A was twice as important? or what if we didnt consider criterion B? This allows decisionmakers to determine the robustness of a given course of action and understand the key drivers of the decision. Once this
Figure 1: Defining relevant criteria, objectives or goals.
23
A N A LY T I C S
has been accomplished, the cost (dollars, time or even people) of each option can be added and any constraint in resources. The result is an optimized allocation of resources among the options using a cost/benefit analysis. By bringing together the available data along with the judgments of decision-makers in a structured way, AHP helps decision-makers see an otherwise elusive, holistic picture of how their goals and understanding of the problem fit their choices. Using the principles of comparative judgment, AHP allows each decisionmaker to express his or her individual priorities amongst the key criteria that form the mental model for their decision and combines each decision makers judgments with those of others. AHP facilitates a greater understanding, within a sense of fair process and participation with greater transparency and traceability of the preferences of groups faced with complex decisions.
CASE STUDY: A COMPARISON OF FINANCIAL BASED CAPITAL ALLOCATION DECISIONS WITH AHP
No matter how diligently organizations attempt to risk adjust the value of their potential projects or programs, the social dynamics of decision-making
can make estimates optimistic. In one study of 1,800 organizations, 80 percent couldnt reach 5.5 percent annual revenue growth, falling far short of their strategic plan goals. While financial forecasting methods attempt to compensate for these effects with risk adjustments, the inputs to financial models are often limited to optimistic estimates provided by project champions or sponsors influenced by the optimistic biases. Often, the need or desire for projects may cause a systemic bias that inflates the estimates of champions and sponsors. These blind spots are seemingly inevitable. Yet, if we are to believe what research tells us, some of this systemic bias can be offset by crowd-sourcing the evaluations of experts across an organization using AHP as a framework for synthesizing their insights. A surprising insight came to light in work done by a company in the pharmaceutical and life sciences industry. While this organization was regularly relying upon risk-adjusted revenues for their product lines, these risk adjustments were provided to the finance department by sponsor and champion organizations proposing their projects. The organization was able to classify projects using a risk scorecard that
looked at how new elements of both the supply chain and technology competencies of the organization fit the project. They found strong correlations between the newness of the project offerings and how likely they were to be completed. They also found correlations between these new, different projects and how likely they would be to meet risk adjusted revenue forecasts. It turned out that a secondary risk adjustment would be needed to further adjust expectations based on these probability of success assessments. However, getting these into the financial analysis proved challenging. AHP was used to blend together both the quantitative strength of the risk adjusted revenue forecast with the qualitative probability of success criteria. The organization was able to compare expected
revenues using the purely financial approach vs. the AHP approach. One hundred thousand simulations comparing the largest forecasted revenue portfolio against the AHP recommended prioritization showed that organizations can consistently expect a greater return on sales using a crowd-sourced, collaborative decision-making approach over purely financial methods, which have hidden assumptions and biases. Using the same set of assumptions for the likelihood of completion and revenue generation for projects, two portfolios were selected. One portfolio was chosen from a list of largest-to-smallest revenue forecast projects, from the top down, until available resources to complete them ran out. The second portfolio was chosen using the cost/ benefit approach of AHP, which included
Figure 4: Forecasted revenue portfolio vs. AHP recommended prioritization.
weighted criteria that combined financial metrics with qualitative probability of success estimates. The AHP-selected portfolio consistently returned as valuable a portfolio as the revenue forecasted approach and 80 percent of the time outperformed it. When the AHP method did outperform the purely financial method, it did so by 20 percent.
CONCLUSION
Figure 3: Lower risk, greater percentage of return.
The Analytic Hierarchy Process meets the challenge of bringing together business intelligence and data mining technologies with the social aspects of human decisionmaking. AHP creates a true picture of the landscape of a decision. It is a means to help decision-makers articulate important decision data points, and combines them with intangible factors that may be difficult
to measure but can be expressed and categorized. By narrowing in on the measures of interest, decision-makers can eliminate the sense of data overload from todays information tidal wave and acquire those pieces of information that are believed to be most relevant. Group decision-making is difficult. Creating a framework and process to deal with it can make the process less error-prone. AHP assists decision-makers in creating a real context for their choices that adequately weights the dimensions they trust or believe influence their path to success. It is the only chance organizations have to safely navigate the rising tide of information.
Kevin Connor is vice president of Decision Lens Solutions Group. He can be reached at kconnor@ decisionlens.com.
25
A N A LY T I C S
CAREER BUILDER
Certied Analytics Professional

INFORMS prepares to launch first-of-its-kind program.
BY SCOTT NESTLER, JACK LEVIS AND BILL KLIMACK (LEFT TO RIGHT)

The Institute for Operations Research and the Management Sciences (INFORMS), publishers of Analytics magazine, is launching an analytics certification program (Certified Analytics Professional or CAP) designed to enableanalytics professionals (and their employers) to have confidence that a person willbring acore set of analytics skills to a project team. Open to all qualified
analysts, the first-of-its-kind certification will consist of a standardized test, as well as a review of the candidates resume or work portfolio. The program was outlined earlier this year in the March/April issue of Analytics. Following is an update on the program, with the first exams scheduled for the 2013 INFORMS Conference on Business Analytics & Operations Research in San Antonio, Texas, April 7-9, 2013, followed by a second opportunity at
the 2013 INFORMS Annual Meeting, Oct. 6-9, 2013 in Minneapolis. INFORMS has published the eligibility criteria for the CAP certification at INFORMS Online including the following: BA/BS (or higher) degree, and at least five years of analytics workrelated experience for BA/BS holder in related area, or at least three years of analytics workrelated experience for MA/MS (or
higher) holder in related area, or at least seven years of experience for those with BA/BS in unrelated area, and verification of soft skills/provision of business value by employer.
JOB TASK ANALYSIS
In general, a Job Task Analysis (JTA) is a comprehensive description of the duties and responsibilities of
26
A N A LY T I C S
C E R T I F I E D A N A LY T I C S P R O F E S S I O N A L
a profession, occupation or specialty area; our approach consists of four elements: 1) domains of practice, 2) tasks performed, 3) knowledge required for effective performance on the job, and 4) domain weights that account for the importance of and frequency with which the tasks are performed. More specifically, the JTA for the CAP program can be viewed as an outline of a partial body of knowledge, as it represents a delineation of common or typical tasks performed and knowledge applied by analytics professionals, grouped
INFORMS Analytics Credentialing Job Task Analysis Working Group

Jeff Camm (Univ. of Cincinnati) Arnie Greenland (IBM Global Bus. Serv.) Bill Klimack (Chevron) * # Jack Levis (UPS) * # Daymond Ling (Canadian Imperial Bank of Commerce) N Freeman Marvin (Innovative Decisions Inc.) Scott Nestler (Naval Postgraduate School) # Jerry Oglesby (SAS) Michael Rappa (NC State / Inst. Adv. Analytics) # Tim Rey (Dow Chemical) Rita Salam (Gartner) N Sam Savage (Stanford / Vector Economics) * - INFORMS Board Member # - Credentialing Task Force Member N Non-Member of INFORMS
together in a hierarchical domain structure. In the course of analytics work, these tasks may be performed multiple times with modifications based on data, findings and results, as part of ongoing feedback loops that are routinely a part of practice. The JTA serves as the test blue print for exam development and links what is done on the job with what is measured by the certification examination. This linkage is necessary to establish a valid, practice-related ex amination. It is important to realize that the JTA is a dynamic document that will change in the future to reflect best practices and changes in the analytics profession. The JTA outlined in this article was developed by the INFORMS Analytics Credentialing Job Task Analysis Working Group, comprised of 12 subject matter experts (SMEs) (see box) who are: highly regarded in their field; diverse in geography, sector (public-private), organization type (e.g., large companies-smaller consulting firms, practice-academia, etc.) and application area (e.g., finance, logistics, software, consumer goods, etc.); and representative of the descriptive, predictive and prescriptive segments of analytics. Additionally, the working group contains four members in common with the task force and two INFORMS
directors, helping to ensure continuity with existing governance structures. Additionally, since CAP is designed to attract analytics professionals who are not currently members of INFORMS, the working group contains some nonmembers. In developing the JTA, members of the working group relied upon: their knowledge of practice gained from years of experience, academic program content, corporate job descriptions in analytics and articles from professional and
scholarly publications. As outlined in the earlier update, the JTA Working Group proposed, and the Credentialing Task Force and Board of Directors approved, that the CAP assess to some level of depth across the breadth of knowledge needed in analytics. The evaluation of more detailed knowledge in specific areas or applications will be done later in add-on certifications, pending the successful development and deployment of CAP.
BOSTON | SEP 30 - OCT 4, 2012

Keynote Speakers:
Predictive Analytics World is the business-focused event for predictive analytics professionals, managers and commercial practitioners. This conference delivers case studies, expertise and resources to achieve: Bigger wins: Strengthen the impact of predictive analytics deployment Broader capabilities: Establish new opportunities with predictive analytics Scott Nicholson
Chief Data Scientist Accretive Health (formerly LinkedIn)
15% OFF
use code INFB1215
for INFORMS members
Anne Robinson
Dir. of Supply Chain Strategy & Analytics Verizon Wireless
Robert Jewell
Dir. Global Bus Dev & Partnerships IBM Watson Solutions
www.pawcon.com/boston/2012
A N A LY T I C S
PRODUCED BY
27
The JTA can be viewed as an outline of a partial body of knowledge, as it represents a delineation of common or typical tasks performed and knowledge applied by analytics professionals, grouped together in a hierarchical domain structure.
DOMAINS PROVIDE TOP-LEVEL STRUCTURE
The 36 typical tasks and 16 knowl edge statements (not provided here) in the analytics JTA are organized in seven domains, as listed in Figure 1. Tasks are specific goal-directed work activities or groups of closely related work activities that describe identifiable behaviors, while knowledge is an organized body of information that, when applied, makes possible the competent and effective performance of the work activities described by a task.
Domain
I II III IV V VI VII
Description
Business Problem (Question) Framing Analytics Problem Framing Data Methodology (Approach) Selection Model Building Deployment Life Cycle Management
Weight
15% 17% 22% 15% 16% 9% 6% 100%
discussion and debate that continued until consensus was reached. The weights will be used in the exam construction process to ensure content mixture validity. Within each of these seven domains, the JTA Working Group identified a number of tasks that must be performed by practitioners of analytics. Successful performance of the tasks requires specific knowledge, which is what will be tested. At this time, the supporting knowledge statements are being used to develop items (questions) for the first exam and, as such, are not publicly available. We anticipate releas ing them as part of the CAP Candidate Handbook and at INFORMS Online in October 2012.
JTA VALIDATION
study and registered subscribers of Analytics magazine. Survey participants were asked to perform three tasks in their review of the draft JTA document: 1. Identify those domains, tasks, or knowledge statements they would like to remove, reword, or revise. 2. Suggest new domains, tasks, knowledge or skill statements that they would like to add. 3. Confirm or suggest changes to the weights based on their ranking of importance and frequency. After reviewing the results of the survey, including a thorough report prepared by the certification consultant, the JTA Working Group met again by telephone in February 2012 to clarify and improve the JTA. The agreed upon changes primarily included the addition of examples of concepts and definitions to most of the knowledge statement in order to ensure understandability.
TEST DEVELOPMENT
Figure 1: Domains, descriptions and weights in the JTA.
Figure 1 also shows domain weights, which are based on the SMEs assessments of the importance of tasks and the frequency of their performance. Mean weights of the working group members were used as starting points for
In order to ensure that the JTA Working Group had not missed anything important in the practice of analytics, the JTA and an associated questionnaire were sent out to a random sample of INFORMS members and non-members. More than 200 analytics professionals from various regions of the United States, Europe and Asia/Pacific responded to the survey; approximately three-quarters of these were not INFORMS members. Non-member respondents included previous respondents of the 2011 certification feasibility
In May 2012, a call for question writers went out via the INFORMS subdivision (section and society) officers and also to those who responded to the feasibility study in 2011 saying they wanted to help with development of an analytics certification program. Nearly 50 volunteers, a
28
A N A LY T I C S
The goal of this training was several-fold: to maximize the quality of the test; to ensure the accuracy, fairness and validity of the test; to minimize the measurement error of the test; and most importantly, to minimize errors in the classification of candidates as certified or not.
mix of academics and practitioners, initially responded; most, but not all, are INFORMS members. In early June 2012, a consulting psychometrician provided training to about 30 volunteers on how to write multiple-choice questions for the exam. The goal of this training was several-fold: to maximize the quality of the test; to ensure the accuracy, fairness and validity of the test; to minimize the measurement error of the test; and most importantly, to minimize errors in the classification of candidates as certified or not. Item writers were asked to keep the following general guidelines in mind: 1. Does the question test something relevant and non-trivial? 2. Does the question reflect current best-practice? 3. Is the question stated clearly enough so that the knowledgeable candidate will be able to select the correct choice without undue hesitation? 4. Is the context, setting and content of the question equally appropriate and familiar to all segments of the candidate population, including minority groups?
5. Is the question free of language and/or descriptions that might be offensive to any segment of the candidate population? 6. Is the content of the question free of language, descriptions or terminology that could reinforce common stereotypes concerning any segment of the candidate population? Over a six-week period, this group worked individually to develop multiple-choice questions for the certification exam. In addition to the question, each writer provided a correct answer, three incorrect but plausible answers, a reference or citation, and a suggestion of which domain, task and knowledge statements the question was useful to assess. In the first round, this group provided more than 200 questions, which were reviewed by 12 volunteers in late June 2012. The question writer and reviewer group included four members of the JTA Working Group, who helped to guide the review effort. In conjunction with a certification consultant, the INFORMS Certification Task Force prepared a draft Policies and Procedures (P&P) manual for the CAP program. Further work on an independent governance board for the program continues as well. Additionally, the
INFORMS staff is working on a branding and marketing plan for CAP. The marketing department developed a logo that was recently approved by the Certification Task Force.
FUTURE STEPS
Test development efforts, to include additional question writing, creation and review of two exam forms, and a cut score study, are ongoing. Development of a detailed marketing and communications plan, a candidate handbook, marketing materials and a Website, as well as the back-end accounting and IT systems, are also underway. The INFORMS Certification Task Force recognizes the importance of making additional details available as soon as possible given the proposed administration date of the first certification exam. To find out more, check the CAP page at the INFORMS Website from time to time. Similarly, keep reading OR/MS Today and Analytics magazine, as further updates will be shared there as well.
Scott Nestler is a colonel in the U.S. Army, currently attending the Army War College. Jack Levis is the director of Process Management at UPS and the INFORMS VP for Practice Activities. Bill Klimack is a decision analysis consultant at Chevron and the INFORMS VP for Meetings. All three are members of both the INFORMS Certification Task Force and also the INFORMS Analytics Certification Job Task Analysis Working Group.
Questions about Analytics Certification?

Contact INFORMS at certification@mail.informs.org
29
A N A LY T I C S
Q U A N T I TAT I V E H I S TO R I A N S P E R S P E C T I V E
Predicting the presidential election

Will the 13 keys once again open the door to the White House?
BY DOUG SAMUELSON
As most U.S. readers of Analytics are no doubt painfully aware, we are in the middle of the most expensive presidential election campaign in history, by far. The Obama and Romney campaigns seem likely to raise and spend, combined, more than $3 billion, with much of the money going to an overwhelming blitz of TV ads. What TV time isnt consumed by these ads seems to be filled with commentary and punditry. As a friend of this
reporters remarked, By October, there wont be 15 seconds available on TV for someone to sell soap. But to what effect? According to at least one source, not much. Last summer, with the presidential election a year and a half away, campaigning and punditry were already intense, with many commentators noting Obamas low approval numbers in the polls and pronouncing him highly vulnerable. The most reliable model of presidential elections, however, indicated a
Mitt Romney (left) and Barack Obama: a tight battle for the White House.
30
A N A LY T I C S
E L E C T I O N A N A LY T I C S
The 13 keys to the presidency

Professor Allan Lichtmans 13 keys and his assessment of how they turn: 1. After the midterm election, the incumbent party holds more seats in the U. S. House of Representatives than it did after the preceding midterm election. (FALSE) 2. The incumbent-party nominee gets at least two-thirds of the vote on the first ballot at the nominating convention. (TRUE) 3. The incumbent-party candidate is the sitting president. (TRUE) 4. There is no third-party or independent candidacy that wins at least five percent of the vote. (TRUE) 5. The economy is not in recession during the campaign. (TRUE) 6. Real (constant-dollar) per capita economic growth during the term equals or exceeds mean growth for the preceding two terms. (FALSE) 7. The administration achieves a major policy change during the term, on the order of the New Deal or the first-term Reagan revolution. (TRUE) 8. There has been no major social unrest during the term, sufficient to cause deep concerns about the unraveling of society. (TRUE) 9. There is no broad recognition of a scandal that directly touches the president. (TRUE) 10. There has been no military or foreign policy failure during the term, substantial enough that it appears to undermine Americas national interests significantly or threaten its standing in the world. (TRUE) 11. There has been a military or foreign policy success during the term substantial enough to advance Americas national interests or improve its standing in the world. (TRUE) 12. The incumbent-party candidate is charismatic or is a national hero. (FALSE) 13. The challenger is not charismatic and is not a national hero. (TRUE) If six or more of these statements are false, the incumbent party loses.
different answer. Based on his 13 Keys model, quantitative historian Allan Lichtman called the 2012 election for Obama nearly a year ago, with some caveats about how some keys might change. His forecast hasnt changed, but the amount of money being raised and poured into TV ads, among other factors, is much different from the past. How might the election go this time? How can we tell? Professor Lichtmans assessment: The 13 keys model has been remarkably stable through all kinds of variations that led people to say, This election is different. This record of reliability and the implications of which variables made it into the model still indicate that presidential elections are about governance, not campaigning. The punch line is that most of that money will be wasted. The one thing it will do, especially with all these attack ads, is get people so fed up with all of the political process that it will be hard for whoever wins to govern. OR/MS Today [the membership magazine of INFORMS, www.orms-today.org] may remember Allan Lichtman, professor of history at The American University in Washington, D.C. He was the subject of feature articles in OR/MS Today in 1996, 2000, 2004, 2008 and 2011. More significantly, he has attracted quite a bit of coverage in the news media, his books continue
to sell well, and it is evident that campaign strategists take his model into consideration in their planning. His model deserves to be taken seriously, as it has correctly predicted the popular vote outcome of every U. S. presidential election since 1984, including George H. W. Bushs comeback from nearly 20 percent behind in the polls in 1988, and Al Gores narrow win in 2000. His predictions are based on 13 questions (see box), each with a yes or no answer. Yes answers favor the incumbent
party. If five or fewer answers are no, the incumbent party retains the presidency; if six or more are no, the challenger wins. For this election, Lichtman said the Democrats have lost Key 1 (the 2010 mid-term election was a huge setback), Key 6 (long-term economic growth) and Key 12 (the incumbent-party candidate is not very charismatic or a national hero.) This leaves the Republicans three keys short of what they need, with the Democrats holding 10 keys.
Deadline for Applications is Monday, October 29, 2012.
call for applications 2013
The UPS George D. Smith Prize is created in the spirit of strengthening ties between industry and the schools of higher education that graduate young practitioners of operations research. INFORMS, with the help of CPMS, will award the prize to an academic department or program for effective and innovative preparation of students to be good practitioners of operations research, management science, or analytics. The UPS George D. Smith Prize is named in honor of the late UPS Chief Executive Officer who was a champion of operations researchers at a leading Fortune 500 corporation. UPS has generously underwritten the award in his memory. The prize will be awarded to an academic program or department. It will include a trophy and a $10,000 cash award. The UPS George D. Smith Prize will be announced at the 2013 Edelman Gala at the INFORMS Conference on Business Analytics and Operations Reasearch.
For more information, questions can be sent to Bob Smith, 2013 Smith Prize chair at ing ties hen dsmith@monmouth.edu. t g en academia str industry http://www.informs.org/smithprize
ups george d smith prize

31
A N A LY T I C S
When Professor Lichtman made his forecast in 2010, some keys were still subject to change, and the current call on some of them deserves comment: If enough elements of governance had gone wrong, a serious challenge either within the party or from a third party could have emerged. Neither did. The economy could have slid back into recession. As of August, however, even the Republicans ads were criticizing the Obama administrations for the slowest recovery since the Depression, a gloomy situation but not a recession. The administrations big policy change, the Affordable Care Act, could have been struck down by the Supreme Court and may still be widely unpopular, but past instances, such as FDRs New Deal or Lyndon Johnsons Great Society, suggest that the ability to enact change, not the way that change plays out, is what turns this key. Major social unrest could have erupted either the Occupy demonstrations or some right-wing activities, such as the privatesector border patrols in Arizona or the controversial protests by the Westboro Baptist Church, could
have escalated into widespread turmoil. 1968 was the last time this key turned, after things were fairly calm in the summer of 1967, so the key can turn quickly. But it didnt. There are always plenty of opportunities for major military or foreign policy failures. So far (as of mid-August), none have occurred. (The key doesnt turn because of strongly critical opinions by a few people, even very knowledgeable people, about certain polices and actions. Broad public perception is what counts.) It was not yet clear a year ago that the killing of Osama bin Laden and the consequent disruption of Al Qaida would be perceived as a major success; now, along with the toppling of Qaddafi in Libya, it does seem to be bringing President Obama substantial credit. Some commentators have also noted that losing control of the House in 2010 may actually help Obama in 2012 because it spreads the blame and gives him someone elses record to run against. Choosing Paul Ryan, chairman of the House Budget Committee, as the Republican vice-presidential nominee intensifies this effect. Losing control of
WIN FRIENDS & INFLUENCE PEOPLE WITH FORESIGHT
Impress your clients and colleagues with your understanding of the latest developments in forecasting research. Dazzle them with anecdotes of real world forecasting successes, pulled straight from our pages. Rally your organization around a common goal of increased efficiency and profitability, supported by your improved forecasts and business planning.
Foresight: The International Journal of Applied Forecasting is a practical guide that relates to business forecasting like no other professional journal can. Four times each year, Foresights pages are packed with articles, reviews, and opinions that showcase the best thinking and writing in the field of forecasting. Visit forecasters.org/foresight to learn more and subscribe or renew today.
FORESIGHT Do you have it? Do you want it? Get it here:

www.forecasters.org/ foresight/subscribe
The 13 keys model has been remarkably stable through all kinds of variations that led people to say, This election is different. This record of reliability and the implications of which variables made it into the model still indicate that presidential elections are about governance, not campaigning. The punch line is that most of that money will be wasted. Allan Lichtman
both houses of Congress in 1994 did not badly damage Bill Clintons re-election prospects in 1996.
THE SCIENCE BEHIND IT

It is worth re-emphasizing that the 13 keys model is based on a statistical pattern recognition algorithm implemented by Russian seismologist Volodia KeilisBorok. In English-language terminology, the technique most closely resembles kernel discriminant function analysis. The idea is to choose the variable that produces the biggest improvement in prediction, then the variable that adds the most improvement, and so on. Lichtman and Keilis-Borok considered many other variables that didnt have much effect: the challenging partys nomination contest, adverse reports on candidates health, running mates, endorsements and massive ad campaigns, among others. The emergence of these 13 variables implies that governance is more important than campaign characteristics. Political consultants hate the Keys, Lichtman says. They keep telling me, Give us something we can influence! But thats not what the model indicates. Of course, he adds, this does not mean that a candidate favored by the keys could simply go home and await the results. Clearly the model assumes the
usual sorts of campaign activities by both sides. Given that, however, the model also implies that much more of the same wont change the outcome. It is also important to note that while massive spending on media blitzes and local organizations may not affect the national popular vote much, it certainly can tip the result in a few closely contested swing states and in Senate and House races, and has substantially affected the nomination primaries and caucuses. The keys model does not address these effects. This models success also underscores the unimportance of poll results this far ahead of the election. Lichtman declares bluntly, polls more than two months before the election are meaningless. Indeed, as he points out, George H. W. Bush trailed Michael Dukakis by 17 points three months before the 1988 election, and Gore trailed George W. Bush in every poll up to the weekend before the 2000 election, and then only edged ahead in one, the Zogby poll. But Gore won the popular vote, as the model predicted.
OTHER THINGS TO WATCH
So here are some things that do matter: Voter access, encouragement and discouragement. In several key states, notably Ohio, Pennsylvania and Virginia, the Republicans are making concerted efforts to tighten requirements, especially with regard to identification documents, for both registration and voting. These restrictions are likely to reduce numbers of votes mostly from poor people, who have problems with transportation to the registrars and to the polls, and from older senior citizens, who more often have expired but not current drivers licenses. Democrats, in turn, typically organize registration drives and offer rides to the polls in areas they expect to have large numbers of Democratic supporters who are less likely than most to vote unless assisted and encouraged. There are several active court cases about how to achieve the appropriate balance between preventing fraud and facilitating access. Differences in access to voting. In Ohio in 2004 and 2006, allocations of voting machines to precincts and provisions for replacing malfunctioning machines contributed to long waiting times, mostly in poor and minority-heavy urban areas. Readers of OR/MS Today a few years ago read about operations research-based
Even if the popular vote is in little doubt, as the keys model implies, the electoral vote is still in play, and many Senate and House races are hotly contested, as well.
33
A N A LY T I C S
efforts to improve the process; the concern continues, however, in Ohio and elsewhere. Vote counting rules. In addition to his work on the keys model, Allan Lichtman has testified as an expert witness in more than 75 cases of alleged wrongdoing in counting votes. In Florida in 2000, he found the way the rules for counting ballots were applied was a critical factor: Ballots with a hole punched for a candidate and the same candidates name written in were disqualified as double votes, even though the voters intent was clear. This rule resulted in the disqualification of more than 120,000 ballots, disproportionately from black voters. If black voters ballots had been rejected at the same rate as white voters ballots, he concluded, there would have been 50,000 more black votes statewide. Assuming the disqualified votes would have split similarly to the black votes that did get counted, this would have tipped the election to Gore handily. Clearly, counting rules are a continuing issue that will command attention. Controversy over the fairness and accuracy of vote counting can also produce lasting discontent with the whole political process, as, for example, with the dispute over the 2004 presidential election in Ohio.
Ethnic blocs. In particular, Latino voters look like the swing bloc in several key states where the races look close: Florida, Ohio, Colorado, New Mexico and Nevada are most frequently mentioned by commentators on this topic. Targeting appeals to selected groups is a well-known and long-standing (more than 50 years old) campaign method that has had considerable effect, particularly in more localized elections, such as those for House districts. In some cases, however, targeting also raises questions of propriety and of possible undermining of the integrity of the political process. Negative messages, either localized or on a large scale, risk a phenomenon known in sales as market saturation: people simply get disgusted with both sides. There is a stronger, more specific backlash against negative messages subsequently proven to be false. Base vs. center. Conventional wisdom among political pros is run to the base for the nomination and to the center for the general election. This means that in October, the leading candidate spends more and more time in swing states, while the trailing candidate has to keep shoring up his base. Seen in this context, Governor Romneys choice of a running mate more conservative than he is, and from a geographic region where he would expect to
be strong already (his father was a popular governor of Michigan), suggests that he and his senior advisors may already be worried about his prospects. The effects of massive spending. Big money spent on media blitzes and on organizational activity may not affect the national popular vote by much, but it can and does influence primaries, Senate and House races and the outcomes in a few closely contested swing states. Beyond
a certain point, however, big spending could also create a backlash from people over-saturated with the ads, phone calls and house-to-house canvassers. Tracking how much is enough and how much is too much will be an interesting challenge. The self-serving professional class. One reason candidates keep raising and spending more and more money, for what the keys model indicates is little benefit, is that all the fund-raisers, political
http://computing.society.informs.org
Computing Society
Interested in computer science, artificial intelligence, and their relationship to O.R.? Join the INFORMS Computing Society. PURPOSE
Keeps members abreast of useful developments in computer science, artificial intelligence, and O.R. Provides valuable contacts and visibility through conferences, committees, and social media Contributes to the rigor of research and practice by providing professional development opportunities
JOIN COMPUTING SOCIETY
MEMBER BENEFITS
AWARDS ICS Prize for best English-language paper Student Paper Award Harvey J. Greenberg Award for Service PUBLICATIONS INFORMS Journal on Computing (included with INFORMS membership) Operations Research, Computing, and Homeland Defense MEETINGS Biennial Computing Society Conference Sponsored sessions at INFORMS Annual Meeting COMMUNICATIONS Newsletter published twice per year, features survey articles, meeting information, people in the news, & upcoming events Electronic mailing list, Blog, LinkedIn Group, Twitter
Joining Computing Society (or any subdivision) is easy, and can be done through your profile on the INFORMS website: Go to the INFORMS website (http://www.informs.org) and click on the Member Login link at top right. Log in. If you don't have login information, click on the link "Please send me my login info." Once you are logged in, you will be redirected to the home page. Click on the "Self-Service Center" link at top. Select "Renew or Add Items for 20XX". Scroll down this page to add/edit your community memberships.
STRATEGIC PARTNER
COIN-OR The Computational Infrastructure for Operations Research (COIN-OR) project is an initiative to spur the development of open-source software for the operations research community.
A N A LY T I C S
consultants and pundits keep telling them they have to. This does generate considerable benefits for the fund-raisers, political consultants and pundits. Eugene Burdick, a political scientist and novelist, pointed out most of the applicable ethical issues in his 1956 and 1964 novels. Naturally, these people are also among the first and most committed to argue against any legislation limiting the money or its uses in advertising, just as the major manufacturers of voting machines are among the first and most committed to raise all kinds of creative objections to any strict standards on the reliability and tamperresistance of voting machines.
CONCLUSIONS
of political preferences, would do well to learn about the analytical methods and issues that contribute to improving elections and keeping them credible.
Doug Samuelson (samuelsondoug@yahoo.com), senior INFORMS member and Treasurer, CPMS, is a frequent contributor to Analytics and OR/MS Today, is president of InfoLogix, Inc., a consulting company in Annandale, Va. He worked as a paid campaign staffer in a U. S. Senate campaign in Nevada in 1970, as a county coordinator in a gubernatorial campaign and targeting analyst for a local campaign in California in 1974 and as a Federal Civil Service policy analyst from 1975 to 1982. REFERENCES
1. Burdick, Eugene, 1956, The Ninth Wave, Houghton Mifflin. 2. Burdick, Eugene, 1964, The 480, McGraw-Hill. 3. Csetlar, Maralee, July 12, 2010, Scholars 13 Keys Predict Another Obama Win, www.american. edu/media/news/20100712_Lichtman_Predicts_ Obama_Wins_Reelection_2012.cfm 4. Kyle, Susan, Samuelson, D., Scheuren, F. and Vincinanza, N., Winter 2007, Explaining the Differences Between Official Votes and Exit Polls in the 2004 Presidential Election, Chance. 5. Lichtman, Allan J., 2008, The Keys to the White House: A Surefire Guide to Predicting the Next President, 2008 Edition, Rowman & Littlefield Publishers, Lanham, Md. 6. Lichtman, Allan, January 2003, What Really Happened in Floridas 2000 Presidential Election, Journal of Legal Studies, Vol. 32, p. 221. 7. Lichtman, A. J., and Keilis-Borok, V. I., 1981, Pattern Recognition Applied to Presidential Elections in the United States, 1860-1980: Role of Integral Social, Economic and Political Traits, Proceedings of the National Academy of Science, Vol. 78, No. 11, pp. 7.230-7.234. 8. Samuelson, Doug, June 2011, Election 2012: The 13 Keys to the White House, OR/MS Today. 9. Samuelson, Doug, Ted Allen and Michael Bernshteyn, December 2007, The Right Not to Wait: Forecasting and Simulation Reduce Waiting Times to Vote, OR/MS Today.
While political methods and tactics continue to change, some factors seem fairly reliable over the long term in enabling us to predict who will win. Barring major new adverse events, these factors favor President Obamas re-election, at least in the popular vote, with the massive infusions of cash into TV advertising having little practical effect other than to annoy most voters. Issues of voter access and bloc voting could substantially affect the electoral vote, however, and potentially cast doubt on the fairness of the election. OR/MS analysts, regardless
C O R P O R AT E P R O F I L E
How analytics enhance the guest experience at Walt Disney World

Forecasting serves as the analytical foundation for operations planning at the resort. It all starts with the park attendance forecast, which lays out the expected attendance at each park.
In the mid-1960s, Walt Disneys dream was to build a family resort destination like no other. The dream became a reality when Walt Disney World Resort opened in 1971, featuring Magic Kingdom as the centerpiece. Today, millions of guests visit the Walt Disney World Resort each year to experience the worldclass theme parks and hotels. In addition to the four theme parks, there is a never-ending list of activities to enjoy, such as shopping at Downtown Disney, playing golf at one of five challenging courses or plunging down a slide at one of our two water parks. When guests get hungry, more than 300 places to dine await them across the resort, ranging from fast casual to fine dining. Guests are often surprised (and delighted) to discover that the resort has more sommeliers than any other company in the world! In all, more than 60,000 cast members guide guests through these experiences, delivering legendary service and creating memories that last a lifetime. Guests vacationing usually focus on the entertainment and attractions the resort has
to offer. What most guests dont see is the careful planning taking place behind the scenes to run the operation smoothly. In this article, well give you a peek at the role analytics plays in ensuring the guest experience is maximized.
IMPROVING GUEST EXPERIENCE THROUGH FORECASTING
BY PETE BUCZKOWSKI (LEFT) AND HAI CHU

36 |
Forecasting serves as the analytical foundation for operations planning at the Resort. It all starts with the park attendance forecast, which lays out the expected attendance at each park. These predictions are strongly considered when setting park hours and performing other strategic planning. More granular forecasts are required for each individual area, such as guest arrivals at the hotel front desks. The company recently launched a new labor demand planning system, which generates transaction forecasts for every 15-minute period at many locations throughout the property, including park entry turnstiles, quick-service restaurants and merchandise locations. These forecasts help the resort plan labor effectively to ensure guest service standards are met.
A unique side view of the iconic Cinderella Castle, centerpiece of the Magic Kingdom theme park.
The resorts costuming operation is the largest in the world, with more than a million costumes in inventory at Walt Disney World alone. Not only does total garment demand vary by season, but the distribution of individual garment sizes also changes as the cast mix evolves over time. Unlike a traditional inventory system, garments are laundered and placed back on the shelf after cast members return the garments. The recycling nature of the garment inventory is further complicated by occasional garment retires. This dynamic environment requires sophisticated forecasting models to have enough garments to meet cast member demand while not overspending on garments that sit on a shelf.
A N A LY T I C S
Another innovative way the resort uses forecasting is for attraction wait times. The most popular attractions utilize Disneys FASTPASS system a unique virtual queueing system that allows guests to receive a ticket with a designated one-hour window of time when they can return and skip the regular line. The virtual queue Guests enjoy playing in the new Haunted Mansion interactive poses an extra challenge to queue. forecasting the standby wait time since many FASTPASS From a central command guests will return before the guest encenter underneath tering the standby queue reaches the the Magic Kingdom, front of the line. From a central command center unforecasting models are executed every 5-10 derneath the Magic Kingdom, forecasting models are executed every 5-10 minutes to project the minutes to project the return patterns of return patterns of our our FASTPASS guests based on a vaFASTPASS guests based riety of factors, including entertainment on a variety of factors, schedules and the number of FASTPASS including entertainment tickets that have been distributed. These schedules and the number forecasts are posted at the front of the atof FASTPASS tickets that tractions to help guests choose whether have been distributed. to enter the line, take a FASTPASS ticket or return to the attraction later in the day. As an added bonus, these projected wait times are also available on Disneys Mobile Magic smart phone app, which
shares real-time information about the parks throughout the day.

MAKING SMARTER DECISIONS
While accurate forecasting is important, the results are only valuable when utilized to make smarter decisions. As Disney analysts understand more about guests and their preferences, they can utilize analytics to customize offerings and experiences that better match resort guests desires. As one example, data
mining is used to understand what vacation packages are most appealing to different types of guests. This analysis, coupled with optimization models, allows the companys Web site and our call center agents to present offers that provide a more customized vacation planning experience. The resorts table-service restau rant operation leverages optimization in unique ways. Statistical analyses helps the company understand the patterns
SEPT 17-18
WASHINGTON DC
The first vendor-neutral analytics conference for government

Predictive Analytics World for Government (PAWgov.com) is a practically-focused conference that highlights case studies of how government agencies are currently using data analytics to solve real world problems.
for INFORMS members

use code INFB1215
PRODUCED BY
15% OFF
www.predictiveanalyticsworld.com/gov/2012
A N A LY T I C S | S E P T E M B E R / O C T O B E R 2 012
CONNECTING THE TEXT ANALYTICS ECOSYSTEM
INFORMS200
WHEN YOU REGISTER AND SAVE $200!
QUOTE
around party sizes, arrival times and table turn times. This knowledge is incorporated into mathematical models that determine the right mix of tables to best meet guest demand. Another set of models helps develop inven tory templates that embrace the stochastic nature of the operation while maximizing the utilization of the res taurant. A recent update made these models dynamic in nature accounting for the bookings already made while considering the projected reservations up until the actual day arrives. Disney also applies analytics to streamline back-of-house operations. An on-site textiles facility handles nearly 300,000 pounds of laundry every day, servicing costumes from across the operation as well as linens from resort hotels. The resort leverages computer simulation to recreate the facility in a virtual environment. Simulation offers many benefits prior to making physical changes, including identifying potential bottlenecks and testing new concepts or designs that can increase the overall capacity of the facility. The resort has
also utilized simulations in many other operational settings, including attractions, hotel front desks, restaurants and warehousing facilities. One of the resorts newest innovations is the creation of interactive queues, which keep guests entertained while they wait in line for an attraction. The most recently opened queue is at The Haunted Mansion. Prior to entering the attraction, guests will find frightful fun in a new courtyard queue that brings them into an expanded, interactive graveyard featuring a ghoulish organ, a wall of unusual musical instruments guests can orchestrate, a chance to complete a rhyme with the Decomposing Composer and photo opportunities with favorite spirits from the beloved attraction. This new concept changes the way guests navigate through the queue instead of progressing in a linear fashion, they often gravitate to one of the interactive elements found in the queue. The resort is beginning to use agent-based simulation to model these new guest behaviors and look to incorporate the findings into future queue designs.
10th Text Analytics Summit - West

Two-Day Text Analysis Conference | November 13-14th | Hotel Kabuki | San Francisco, CA
EXTRACT CRITICAL INSIGHTS FROM UNSTRUCTURED DATA AND MAGNIFY THE BUSINESS IMPACT GAINED FROM TEXT ANALYTICS
Improve the Business Decision Making Process: Learn why text analytics is the gateway to better decision making and increased productivity Capitalize on Social Media Analytics: Employ social media mining and analysis for competitive intelligence, brand management, and customer engagement Take Advantage of the Most Cutting-Edge Technologies: Use the newest advancements in sentiment analysis, semantics, information retrieval and the cloud to stay ahead of the competition Optimize the Customer Experience Process: See how customer experience analytics uncovers actionable insights into consumer behavior that you have never had access to before Utilize Multi-Channel Analytics: Enhance your big-picture analytics strategy to fully grasp the Voice of the Customer and perform superior research and data analysis PLUS Exclusive case studies from top end user organizations: Get real life examples of how end users have efficiently integrated text analytics into their day-to-day operation
PLATINUM SPONSOR: BRONZE SPONSOR:
Hear From More Than 30 World-Class Text Analytics Experts

Anmol Bhasin Senior Engineering Manager LinkedIn
Ashok Srivastava Principal Scientist NASA
Michael Skinner Principal Patent Manager Intel
Evgeniy Gabrilovich Senior Staff Research Scientist Google
Judy Pastor Principal, Operations Research American Airlines
Request a no-obligation INFORMS Member Benefits Packet

For more information, visit: http://www.informs.org/Membership
Brian Eoff Data Scientist Bitly
38
FOR FULL EVENT DETAILS GO TO www.textanalyticsnews.com/west1
AN EMERGING ANALYTICAL CULTURE
Over the past decade, analytics at Disney have emerged as a powerful means to bring even more magic to its resort guests. Numerous departments across the company are leveraging enhanced tool sets to help solve its most challenging business problems, including forecasting, marketing, industrial engineering, revenue management and transportation, to name just a few. One result of this growth is the annual Disney Analytics & Optimization Summit. Taking place in the fall, it brings together cast members who have analytical roles throughout the company with other external O.R. professionals. Last year, nearly 300 cast members attended the summit a true testament to the volume of analytical work being accomplished and how it is highly valued throughout the company. While much of this work originated at Walt Disney World, the groups are also taking on projects throughout the broader Disney organization. Many of
the successful analytical projects are subsequently implemented at our other parks and resorts around the world. In fact, Disney recently launched two new cruise ships into the Disney Cruise Line family, the Disney Dream and Disney Fantasy. Analytics were utilized throughout the planning process, including organizing the logistics around training the nearly 1,700 new crew members and planning crew rotations on and off the ship. Disney also continues to look for ways to expand its analytical best practices to other segments of the company, including ABC, ESPN, and Studios divisions. With more data becoming available in an ever-changing landscape, Disneys need for analytics will continue to grow in the future.
Pete Buczkowski (peter.s.buczkowski@disney. com), senior INFORMS member and member of INFORMS Roundtable, is the manager of the Advanced Analytics team inside the Disney Industrial Engineering department, providing analytical insight to solve operational challenges at the theme parks and resorts. Hai Chu (hai.d.chu@disney.com) is the director of Decision Science and Support within the Revenue Management & Analytics department. His team builds mathematical models to optimize decision-making across the entire Walt Disney Company.
April 7-9, 2013

San Antonio, TX
Grand Hyatt San Antonio
Present a talk, present a poster, at the premier Analytics Conference and save 35%+ on conference registration.
Present your work at the conference known for real-world analytics and applied O.R. Case studies, best practice examples, and academic research with a practitioner orientation are all welcome. Selected presenters receive a 35-40% discount off regular rates. Open call for presentations, with special focus on: Big data analytics Supply chain management & logistics Marketing analytics Forecasting & risk management Plus all other areas and topics within the business analytics and O.R. arena. Thanks to our Sponsor
For submission information:

http://meetings.informs.org/analytics2013
CONFERENCE PREVIEW
Winter Simulation Conference

Premier international forum set for Dec. 9-12 in Berlin.
WSC offers an invaluable educational opportunity for novices and experts alike, with a large segment of each program devoted to introductory and advanced tutorials.
BY OLIVER ROSE
The Winter Simulation Conference (WSC) has been the premier international forum for disseminating recent advances in the field of system simulation for more than 40 years. The longestrunning conference devoted to simulation as a discipline, this years WSC will be held in Berlin, Germany, Dec. 9-12, at the InterContinental Berlin, located in the heart of the city. The Berlin Zoo and Berlins largest park, the Tiergarten, are just around the corner from the hotel, which is surrounded by the infrastructure of a metropolis and is easy to reach by plane, rail or car.) The theme for WSC 2012 is WSC Goes Europe. Simulation is widely applied across a diverse range of endeavors. This is reflected in WSC 2012, which includes tracks devoted to manufacturing, semiconductor fabrication, transportation and logistics, healthcare, social sciences and organization, construction, military, computer/ communication networks, environmental sciences, project management, education and gaming. Moreover, WSC offers an invaluable educational opportunity for novices and experts alike, with a
Brandenburg Gate.
large segment of each program devoted to introductory and advanced tutorials. Tutorials are carefully designed to address the needs of simulation professionals at all levels of expertise and are presented by prominent individuals in the field. Of particular interest to virtually all attendees are the software tutorials and the exhibits by software vendors, which cover a broad range of commercial simulation products and services. Rounding out the attractions are poster
sessions, a Ph.D. colloquium and several social gatherings, as well as meetings of numerous professional societies and users groups. All of these events give attendees the opportunity to get acquainted and to become involved in the ongoing activities of the international simulation community. Stefan Rahmstorf of the Potsdam Institute for Climate Impact Research in Germany and Thomas A. Henzinger, president of the
40
A N A LY T I C S
CONFERENCE PREVIEW
WSC is designed for professionals at several levels those who want to apply science as well as art to quantitative business problems.
Berlin Zoo.

Institute of Science and Technology (IST) in Austria, will deliver this years keynote presentations. In 1999, Rahmstorf was awarded the $1 million Centennial Fellowship Award of the U.S.-based James S. McDonnell Foundation. Since 2000 he has taught physics of the oceans as a professor at Potsdam University. He is a member of the Academia Europaea and of the German Advisory Council on Global Change. He has published more than 60 scientific papers (14 papers in nature and science), including Our Threatened Oceans (2009, with Katherine Richardson) and The Climate Crisis (2010, with David Archer).
Henzingers research focuses on modern systems theory, especially models, algorithms and tools for the design and verification of reliable software, hardware and embedded systems. His HyTech tool was the first model checker for mixed discrete-continuous systems. He is an ISI highly cited researcher, a member of Academia Europaea, a member of the German Academy of Sciences (Leopoldina), a corresponding member of the Austrian Academy of Sciences, a Fellow of the ACM and a Fellow of the IEEE. Gianfranco Balbo of the University of Torino (Italy) will present the simulation titan talk. In 1984, together with M. Ajmone-Marsan and G. Conte, Balbo proposed the Generalized Stochastic Petri Net (GSPN) formalism, which, supported by the GreatSPN software package, quickly became one of the most popular modeling languages for the specification and analysis of performance models of computer and communication systems. He has published more than 70 scientific papers and co-authored three books. He is an ISI highly cited researcher, a member of the Academy of Sciences of Torino and a member of the ACM. WSC is designed for professionals at several levels those who want to apply science as well as art to quantitative business problems. It is for analysts to
Berlin Cathedral.
improve their productivity, for managers who lead technical staff and for executives who appreciate and want to learn more about the power behind the techniques and tools of simulation modeling. For more information, see www.wintersim.org. WSC regularly attracts the best of the worlds simulation. Come and experience this gathering of the worlds best simulation researchers, thinkers and teachers, while they share their creativity, teach the basics of simulation and provide advanced state-of-the-art concepts and tutorials in modern day simulation. For more information and updates, visit www.wintersim.org/.
Oliver Rose is the general chair of WSC 2012.
41
A N A LY T I C S
F I V E - M I N U T E A N A LY S T
Train Station
How bad is it to get on the wrong train? Weve carefully bounded the problem such that the only costs are in terms of time.
Two men are traveling in a foreign city with no map and limited knowledge of the language, and, finished for the day, are trying to get back to their lodgings. They have arrived at a light rail station as a train is boarding. Should they board the train in front of them, not knowing if it is the right train, or should they wait for more information? Lets suppose that we have some time five minutes, say to analyze this decision. Before proceeding, lets make some bounding assumptions. First, this is a city light rail, and for simplicity we will assume is similar to the Washington, D.C., Metro. If you miss the train, the next stop is only a few minutes away. Second, lets presume that the trains will be in operation longer than the time scope of the problem, so we are not concerned with being on the last train of the night. We also assume that you are able to change directions without leaving the train system. You took the train earlier in the day, but you didnt necessarily get off at the stop you are currently located. Finally, we assume that you will know with certainty as soon as the train leaves the station if you are headed in the correct direction. If we had no knowledge, we would think guessing will make the right choice 50 percent of the time.
How bad is it to get on the wrong train? Weve carefully bounded the problem such that the only costs are in terms of time. Consider the paths of two travelers who we call Rob and Jeff, currently at Crystal City wanting to head south, but are unsure if Huntington or Mt. Vernon Square is the correct direction. At noon, Rob gets on a train headed toward Mt. Vernon Square while Jeff waits for the next train to Huntington. At 12:01, Rob realizes he has chosen the wrong train. At 12:03, he gets off at Pentagon City, crosses over to the other platform, and at 12:05 boards a train headed back toward Crystal City. At 12:10, he arrives back at Crystal City in time to see Jeff get on the train and sit next to him. This is because the whole time Rob was on the wrong train, he was headed toward the right one. This turns out to be the key to thinking about this problem. Lets make this story concrete with some mathematics. Let be the transit time between train stations A and B. Let be the time until the next train headed from B to A leaves station B. Rob will not be penalized for choosing the wrong train so long as Now we just have to figure out what this expression means to our travelers! Lets proceed by assuming that these times are exponentially distributed, but Id like to explain the assumption. Im doing it for two equally important reasons: 1. It makes the mathematics tractable. 2. It is a better model for reality.
Id like to explain (2) first: We know that the time of train arrivals is continuous and positive, i.e., we cannot get on trains that have already departed the station. This gives us two easy distributions to choose from [1]. However, were assuming that weve ridden on the train once already, and so we may wish to estimate the waiting time based on a single sample. This is a very dodgy business, but we have no choice. If we say that train arrivals are uniformly distributed, then we are implicitly saying that we know the longest time we would ever have to wait. If this were the case, we would know the true upper bound on transit times, and the problem would be much simpler than what is posed here. With regards to (1), we have to keep the math simple due to the five-minute analysis. Theres a nice property of the exponential distribution that is worth remembering, and it is frequently called the race of the exponentials. If are exponentially distributed with parameters , then: .
BY HARRISON SCHRAMM
We may use (2) to conclude that if the amount of time we waited for our train in the morning is approximately equal to the time between stations (not an unreasonable assumption), then if we get on the train before us, we will have made the correct choice with 75 percent probability. A more sophisticated approach, beyond the scope of our current discussion, might use the
42
A N A LY T I C S
About the Competition F I V E - M I N U T E A N A LY S T

The purpose of the competition is to bring forward, recognize, and reward outstanding examples of operations research, management science, and advanced analytics in practice. The client organization that uses the winning work receives a prize citation; the authors of the winning work receive a cash award.
If your hotel is near the city centre and it is early when you are headed toward your hotel, you may presume that the majority of travelers at that time may be headed that way as well, and get on the train if the majority of your fellow platformstanders do as well.
number of persons at the station as an estimator for the time remaining until the next train. If we let N be the number of persons on the platform and be the time remaining until the next train, we would be then interested in finding . Bonus: There is another, simpler approach to making this decision that is also worth mentioning. If your hotel is near the city center and it is early [2] when you are headed toward your hotel, you may presume that the majority of travelers at that time may be headed that way as well, and get on the train if the majority of your fellow platformstanders do as well.
Harrison Schramm (harrison.schramm@gmail.com) is a military instructor in the Operations Research Department at the Naval Postgraduate School in Monterey, Calif. He is an INFORMS member. NOTES
1. This is not to say that we are restricted to these choices; it means that these are the ones that are attractive for this effort. 2. By which we mean early to the local custom. Americans may consider 5 p.m. as early, while Europeans may find anything before 11 p.m. as early.
Entry Requirements
Call for Entries

A $15,000 Competition with a $10,000 First Prize
Application Deadline: October 17, 2012
Send us a document containing a three page summary (a maximum of 1500 words) that describes what you accomplished, in just enough detail to let us judge the appropriateness of your work for the competition. Include a cover page with the title, authors names and affiliations, and a 60-word abstract of the achievement. We require that the document be submitted electronically as a pdf file. Visit the website www.informs.org/edelmanaward for more information. All summaries, abstracts, papers, press kits, and presentations must be submitted in English. All papers submitted by finalists become the property of INFORMS. Transmittal of a cash award to a finalist is contingent on the finalist delivering to INFORMS/CPMS, in a specified form, a paper that meets normal publication standards. The first-prize cash award is also contingent on presentation of the work at the CPMS Plenary Session at the INFORMS October 2013 Annual Meeting in Minneapolis, Minnesota. Award winners must satisfy all requirements and standards established for the competition. Entries should report on a completed practical application and must describe results that had a significant, verifiable, and preferably quantifiable impact on the performance of the client organization. Finalist work will be published in the January-February 2014 issue of Interfaces. Any work you have done in recent years is eligible, unless it has previously been described by a Franz Edelman Award finalist. Previous publication of the work does not disqualify it. Anyone is eligible for the competition except a member of the judging panel.
Key Dates for the Competition

Wednesday, October 17, 2012
Deadline to provide a single pdf document containing a three-page summary (a maximum of 1500 words) of your achievement, and a cover page with a 60-word abstract, and the name, address, phone number, and affiliation of each author. We require that the document be submitted electronically as a pdf file.
Monday, December 17, 2012
Finalists will be selected based on the summaries and the INFORMS/CPMS verification process.
Friday, February 15, 2013
A full written paper in a prescribed format will be due from each finalist along with a publication release and copyright transfer from the client organization, a public relations press kit, and the name of the client management representative who would accept the prize citation.
E-mail Submissions
Please e-mail your submission to: pbell@ivey.ca Peter C. Bell Chair, 2013 Edelman Award Competition Richard Ivey School of Business University of Western Ontario
Watch video of 2012 winner TNT Express accepting the Edeman Award at www.scienceofbetter.org/Edelman.
Monday, April 8, 2013

Each finalist group will give an oral presentation of their work in a special session (which will be recorded) at the INFORMS Conference on Business Analytics and Operations Research in San Antonio, Texas, April 79, 2013. Final judging will begin immediately after the presentations, and the results of the competition will be announced at the conference.
The Pracce Secon of INFORMS
CPMS
43
T H I N K I N G A N A LY T I C A L LY
Lost at sea
Searching for a lost ship at sea is a time sensitive task that requires skill and urgency. Quickly finding a lost ship can mean the difference between life and death. The map in Figure 1 shows a section of ocean divided into 64 cells. Somewhere in this grid, a ship has been lost. Each cell has a number that represents the probability of finding the lost ship when that cell is searched (based on last known position, ocean currents and debris sightings). For example, if you searched cell A1, you would have a 2 percent chance of finding the lost ship there. As the leader of the search and rescue team, your goal is to find the ship with all survivors. Unfortunately, it takes you one day to search a cell, and the lost sailors have only enough food and water to survive for 10 days. This allows you to search a total of 10 cells before the lost sailors perish. You may start your search in any of the 64 cells. You are only allowed to move to adjacent cells (you cannot move diagonally), and you are not allowed to revisit any cells. Add up the percentages in the 10 cells you have searched to get the probability of finding the lost ship.
QUESTION:
OPTIMIZATION
High-Level Modeling
The General Algebraic Modeling System (GAMS) is a high-level modeling system for mathematical programming problems. GAMS is tailored for complex, large-scale modeling applications, and allows you to build large maintainable models that can be adapted quickly to new situations. Models are fully portable from one computer platform to another.
www.gams.com
State-of-the-Art Solvers
GAMS incorporates all major commercial and academic state-of-the-art solution technologies for a broad range of problem types.
GAMS Integrated Developer Environment for editing, debugging, solving models, and viewing data.
Figure 1: Where is the lost ship?
Day-Ahead Scheduling (DAS) Solver

DAS Solver, developed by the Power Systems Laboratory of Aristotle University of Thessaloniki, is an integrated software for the solution of the Day Ahead Scheduling of the Greek wholesale electricity market. It uses advanced Mixed Integer Programming (MIP) modeling under GAMS/CPLEX to formulate and solve the Greek day-ahead electricity market clearing problem, respecting all the provisions of the relevant Electricity Code. It has a user-friendly Microsoft Excel-based Graphical User Interface, providing a variety of graphs and tables of the market clearing results. DAS Solver is used by almost all major electricity market participants in Greece to maximize the benefits of their participation in the day-ahead electricity market.
Europe GAMS Software GmbH Eupener Strasse 135-137 50933 Cologne, Germany phone +49-221-949-9170 fax +49-221-949-9171 mail info@gams.de web http://www.gams.com USA GAMS Development Corporation 1217 Potomac Street, NW Washington, DC 20007, USA
BY JOHN TOCZEK
John Toczek is the manager of Decision Support and Analytics for ARAMARK Corporation in the Global Risk Management group. He earned a bachelors degree in chemical engineering at Drexel University (1996) and a masters degree in operations research from Virginia Commonwealth University (2005). He is an INFORMS member.
ARISTOTLE UNIVERSITY OF THESSALONIKI
What is the greatest probability of finding the lost ship? Send your answer to puzzlor@gmail.com by Nov. 5. The winner, chosen randomly from correct answers, will receive an O.R. The Science of Better T-shirt.
44
phone +1-202-342-0180 fax +1-202-342-0181 mail sales@gams.com web http://www.gams.com
For more information contact bakiana@eng.auth.gr

SeptOct2012 PDF

Diunggah oleh

Informasi Dokumen

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

SeptOct2012 PDF

Diunggah oleh

Hak Cipta:

Format Tersedia

Mudders uncover treasure

The AIMMS PRO Platform A New Revolution In Optimization

No? We have! Check out our AIMMS PRO Platform at:

PETER HORNER, EDITOR peter.horner @ mail.informs.org

XLMINER: VISUAL DATA MINING

DRIVING BETTER BUSINESS DECISIONS

RISK SOLVER PLATFORM

TEACH AND LEARN OPTIMIZATION AND SIMULATION

Tel 775 831 0300 Fax 775 831 0314 info@solver.com

The future of data and analytics

THE MOST FLEXIBLE WAY TO DEPLOY YOUR MODEL

SOLVER SDK PLATFORM

Tel 775 831 0300 Fax 775 831 0314 info@solver.com

Feeling the power

A perfect example of a management policy focused on blindly following prescribed metrics.

Colleagues, columns and catching up

Advertising Opportunities at:

Driving protability with retail analytics

BY SALLY TAYLOR-SHOFF (LEFT) AND SHALINI RAGHAVAN

WHAT PROBLEM DOES ANALYTICS SOLVE FOR RETAILERS?

INFORMS HEALTHCARE 2013

2ND CONFERENCE ON HEALTHCARE Spring 2013 (dates to be announced) Chicago, IL

Help Promote Analytics

high-performance real analytics A game changer.

sas.com/269 for a free white paper

Join the Analytics Section of INFORMS

period before competitors can draft off of their momentum.

How will you stand out from the crowd?

A membership in INFORMS will help!

begin online September 11

visit http://join.informs.org after September 11 to join online

Leveraging Big Data around customers

Request a no-obligation INFORMS Member Benefits Packet

Figure 2: Listening to the customer.

Help Promote Analytics

Figure 3: The insights data can provide are unlimited.

WHATS AN ANALYTICS MAGAZINE READER TO DO AT THE 2012 INFORMS ANNUAL MEETING?

ATTEND AN ANALYTICS SESSION

MEET ANALYTICS SECTION MEMBERS

RECEIVE INDIVIDUAL CONSULTING

Figure 4: The technology stack.

Decision Sciences Institute

Decision Sciences Institute

Taming the data tidal wave

THE INNER WORKINGS OF AHP: MERGING DATA AND EXPERT JUDGMENT

Figure 2: Sensitivity analysis clarifies decision alternatives.

Figure 1: Defining relevant criteria, objectives or goals.

Figure 4: Forecasted revenue portfolio vs. AHP recommended prioritization.

Figure 3: Lower risk, greater percentage of return.

Certied Analytics Professional

BY SCOTT NESTLER, JACK LEVIS AND BILL KLIMACK (LEFT TO RIGHT)

INFORMS Analytics Credentialing Job Task Analysis Working Group

BOSTON | SEP 30 - OCT 4, 2012

for INFORMS members

DOMAINS PROVIDE TOP-LEVEL STRUCTURE

Figure 1: Domains, descriptions and weights in the JTA.

Questions about Analytics Certification?

Predicting the presidential election

The 13 keys to the presidency

Deadline for Applications is Monday, October 29, 2012.

call for applications 2013

ups george d smith prize

WIN FRIENDS & INFLUENCE PEOPLE WITH FORESIGHT