(1)   Google Insights for Search
Hal Varian, Google’s Chief Economist, gave a compelling keynote that made great use of this free tool.  Check it out if you haven’t done so already.  It’s a great way to identify market trends based on search frequency.  The data goes all the way back to 2004, and you can download the results to combine with other data sources.

(2)   It’s not just about the analytics
Congratulations to TNT Express, winners of the 2012 Edelman Award.   My biggest takeaway from their presentation had nothing to do with their network optimization models.  Rather, it was how they built an entire optimization “ecosystem” centered on their GO (Global Optimisation) Academy.

• “(In addition) to adopting OR tools, TNT Express has teamed up with the Tilburg University/TiasNimbas Business School (Netherlands) to create a two-year management development program in transportation network optimization. Called the GO Academy, the training has been delivered to more than 150 managers from TNT since 2008.”

Lots of companies have advanced optimization capabilities, but TNT have added huge value to the organization – and set a great example for other companies — through leadership, education, training and change management.

(3) The role of Analytics in Decision-Making
We’re always looking for better ways to describe what we do and how we add value through Analytics, and I thought a few of the speakers had particularly interesting angles.

• Thomas Olavson of Google described his team’s role as “replacing intuition with data-driven decisions”.
• Colin Kessinger of End-to-End Analytics reinforced the role of leadership.  “Math supports the decision-making process, but the math doesn’t make the decisions”.  Don’t expect an analytical model to give you “the answer”.
• Both Colin and Glenn Wegryn of P&G focused on visualization & interactive decision support.  According to Colin, “This is what gets you a seat at the executive table”, and Glenn provided the evidence when he described the weekly “immersive business reviews” for P&G’s executive team (with a great picture of the huge hi-def screens that surround the conference room)
• Focus on the “so what” or the “why” – not the “what”.  Reports are a dime a dozen.  How are you going to help drive better decisions?
• A neat quote from Glenn Wegryn:  “We bear gifts – in the form of rational business insights”.  (In my words, “Geeks bearing gifts” — OK, that’s a really bad pun…)

(4)   Simpler is Better
The simplest approach that solves your problem is usually the best one.  Several speakers hit on this point in different ways.

• There is “no correlation between analytic complexity and business value”, according to Glenn Bailey of Manheim.
• Chris Fry of Strategic Management Solutions paraphrased Einstein – A model “should be as simple as possible, but no simpler”.
• Before you jump into a big, complex analysis with expensive 3^rd-party software, make sure you really understand the problem you’re trying to solve.  Colin Kessinger advises that if you can’t build and understand a prototype in Excel, you don’t really understand the problem and you’re not likely to be successful with a complicated off-the-shelf software solution.
• Thomas Olavson echoed the sentiment that “small steps lead to big wins”.  He favors analytical building blocks and rapid prototypes, and cautions against looking for “grand unified theories”.

(5)   50 Minutes with the 5 minute analyst
It seems only appropriate to make the 5 Minute Analyst my 5^th highlight.  In a very interactive conference session, Harrison Schramm of the Naval Postgraduate School shared some fun and thought-provoking “toy problems”.  Be sure to check out the 5 Minute Analyst column in Analytics Magazine.  The most recent column is available here.

Please let me know your comments.  It’s been fun being one of your Conference Bloggers and I look forward to continuing these interesting discussions.  Until next time…

Rob

I attended plenty of fascinating talks, from the Intel presentation in the Edelman competition to the talk on analytics in insurance by an IBM professional, and met many wonderful OR professionals, such as the future boss of my doctoral student who is graduating this month.

The conference was extremely well-run and to finish I’d like to take a moment to acknowledge the INFORMS people who manned the registration booth for long hours, and the prize committee members who spent a lot of time evaluating submissions and maintain the high standards of INFORMS – in particular those who participated to multiple committees, such as Susan Albin, Cindy Barnhart, Mike Gorman and Bob Smith (with apologies to the ones I am forgetting). Their work is much appreciated.

Although this year’s accomplishments will be hard to beat, I’m looking forward to next year’s analytics conference already!

If you missed the brief, check out his article in the Analytics online magazine.

http://www.analytics-magazine.org/march-april-2012/542-five-minute-analyst-wikipedias-protest

- The best leaders don’t manage people, policy and systems – they intentionally manage culture.

- The Edelman Finalist met the difficult task of demonstrating how their Analytic/Operations Research techniques improved their task, project, or company.  Getting the right people to either support or utilize their technique was a key point in many of their presentations (I admit to not attending all of them).  That is a statement of culture.

- The Edelman Award winner, TNT Express, demonstrated that they intentionally undertook a plan to infuse their DNA with Operations Research – to move to a new culture.

- Decision Coaches arise as a new profession – some cultures can adapt easier to this than others.

In the end, it is all about culture.  Saving the organization money is good.  Increasing profits is good.  Growing a culture of Analytics / Operations Research is good that endures.

Among the ideas we discussed:

Data cleaning & validation

• Work from the “edges in” to make sure you’ve captured all necessary data
• Compare relevant “sums” to some external data source to make sure nothing is missing
• Look for outliers — data points that don’t look “normal.  Look for min and max values; sort the data, run histograms and scatter charts

Modeling practice

• Standardize cell colors, e.g., input cells blue, decision variables red, outputs green (or whatever color scheme you prefer)
• Maintain integrity of source data; don’t perform ANY operations on the original data set
• Use revision control.  Tools are available that can capture an audit trail, or to track which users have which version of a model.

Error Prevention

• Use Peer Review (best practice from software development)
• Use an auditing tool
• Keep it simple!  If a non-technical person can understand your model, you probably got it right.  Try your spouse or your children!

We hope the attendees enjoyed the session.  Let Rick or me know if you have any follow-up questions or comments.

Check out the SPRIG – Spreadsheet Productivity Research Interest Group.

Rob

These experiences show that our community has indeed embraced the broader analytical message. It wasn’t long ago that it was sufficient to single out a closest analytical area and tack it on to a favorite OR field. As if there was hesitation in stepping too far out of our individual OR comfort zone, as if we felt compelled to expand our tool set, but weren’t sure we really wanted to. Closest depended on who you were, I guess, and what your area of applications was; to some it seemed to be statistics, to others it was data mining, and to others forecasting. But even very recently it felt that there was resistance to the fact that OR is part of a much bigger whole, a whole that is much much bigger than “OR +”. After yesterday I am convinced that we are past that.

Take optimization, for example. Optimization consumes data, forecasts, predictive models, and statistical information. It feeds simulations, what-if analysis, trade-off analysis, and decision support systems. It has to work in a seamless environment, provide answers that can be easily understood, and adapt to changing business needs. It is often subservient to the all-mighty GUI and has to play nice with data systems. It is a cog in a machine, and we get that. Value doesn’t come from a 10% faster MILP solve, it comes from enabling a business user (who knows nothing about cutting planes but knows everything about cutting big checks) to make quick decisions about his supply chain, and to feel good about results and the experience.

But we are still evolving. And to those of us still timidly dipping their toes in the Analytical waters, I say: jump in! You’ll have plenty of opportunities today.

But beyond the big data hype, I noticed another common thread across the presentations I saw yesterday.  As you dig into the mountains and mountains of big data, you often run into unexpected sparcity. As Hal drilled down into the search data, he mentioned that there are search queries that don’t appear very often even in the hundreds of millions of searches that are typed into Google every day. And, despite the power of a petabyte of market basket, Rex acknowledges that they still don’t know quite what to do with new products that have little market history.  So, uncertainty still exists if you know where to look for it.

In fact, characterizing uncertainty may be more important than ever as we face situations where we know some things with near certainty (like 12 oz Coke is almost always the most frequently purchases SKU in a grocery store) and then run into other situations where we know very little (like how well scrapple, my favorite Philadelphia delicacy, would sell in a Southern California grocery).  Big data provides us with the challenge of measuring this uncertainty, incorporating it into prescriptive models and, most importantly, helping managers understand that even with petabytes of data there are still some things we just don’t know.

So first’s thing first….at a minimum I have to post a shout out to my colleagues at Booz Allen Hamilton for winning the Innovation in Analytics Award.  Only in Operations Research would you ever see Booz Allen Hamilton, IBM and the Cleveland Indians in the running for the same award.  Congratulations!

Next up, Congratulations to TNT Express for wining the Edelman! I can’t wait to see their award wining presentation tomorrow.  The

As far as the presentations go, I’d have to say that the presentations I have seen on Track 2 “The Analytics Process” resonate with me the most.  Managing organizational change and building/modifying the right infrastructure has been critical in the success of my projects.  I know its not something we normally want to tackle first, but it has become very apparent to me that nailing down this aspect of an Analytics project is critical to adaption and sustainment of the solutions delivered.   This is why I have always thought that us analytic types need to partner up with strategic policy types and IT/Developer types.  I don’t know of one person that is truly an expert in all aspects of developing and implementing an Analytic solution.  But when partnered up with non-Analytics experts, together, we would be a Problem Solving dream team!