Data and Opinions

I give a lot of talks to our students touting the value of data in business decisions.  The Tepper School prides itself on its analytical approach to business, and data is at the heart of this approach.  Of course, what you do with the data is also pretty important, which brings in statistics and operations research.

For years, I have used the following quote in my talks:

Without data, you’re just another person with an opinion.

That quote works (I sometimes substitute a cruder word for “person”), but some students continue to think “OK, so let’s talk opinions”, which is not quite what I have in mind.  So I think I will change this to a great quote from Jim Barksdale, former CEO of Netscape:

If we have data, let’s look at data. If all we have are opinions, let’s go with mine.

That does the job!

HT: Thanks to @JohnDCook whose tweet reminded me of quote.

 

Operations Research and a Baseball Job

Analytics is getting to be more and more important in sports, and sports teams and leagues are looking to people with analytical skills to fill key roles in their organizations.   The MIT Sports Analytics conference is a big deal, attracting more than 2000 attendees, with an active job placement service.  The MBAs at my own school (the Tepper School) now has a sports analytics club, with a speaker series, case competition and more (including fun things like fantasy sports competitions) and many of these exceptionally bright and ambitious students are eager for jobs in the sports industry.  While some of this may be due to the success of Moneyball, much more of this is due to the fact that computers and decision making have gotten much, much better in the last years, making analytics a key competitive advantage.  And when you get past dashboards and basic data analysis and visualization, you move into using data to make better decisions.  In other words, you move into operations research.

It is clear that many clubs in Major League Baseball get it.  I see it when talking to people with my local team, the Pittsburgh Pirates (a team that I am sure will break .500 any year now!), and I just got a job announcement that shows that the next closest team to me, the Cleveland Indians, get it too.  They are looking for a VP-Technology, but it is clear that they see this as a job involving decision making, not just infrastructure.  From the ad, the primary purpose is:

The Vice President of Technology is responsible for developing, implementing, measuring and maintaining
plans that advance the organization’s achievement of its guiding commitments through enhanced
Baseball Operations and business decision-making tools, increased effectiveness of systems, hardware,
technology infrastructure and improved fan experience through fan-centric technology implementations.

I love the “decision-making tools” in that description.  Sounds just right for an operations research person who also understands technology.

 

So What Correlates with Operations Research?

Google Labs has a new tool called Google Correlate. Google provided some early correlation results during the 2008 flu season when it showed that search count for certain terms (like “flu” presumably) could be used to estimate the prevalence of flu in an area.  This led to Google Flu Trends (it appears that currently only South Africa has many cases of the flu).

You can now play this game on your own data.  Have a time series over the last 9 years or so?  You can enter it into Google Correlate and see what search terms are correlated with the data.

Even easier is just entering a search term:  it will then return other correlated search terms.

If you are going to periodically write in something called “Michael Trick’s Operations Research Blog”, it is clear what to do next:  search on “Michael Trick” (it is required to egotistically search on your own name first, right?).  No dice:  I’m not popular enough to justify a search (sigh…).

But, of course, “operations research” works fine.  What  correlates with that phrase?  Turns out lots of interesting things:  “signal processing”, “information systems”, and … “molecular biology”?  What are the common features on these terms?  Well, they were relatively more common search terms in 2004-2005, relatively flat in the past three years, and have a strong seasonality, corresponding to the start of the academic year (“Hey, I signed up for Operations Research:  what the heck is that?”).  Whether it is operations research, signal processing or molecular biology, it appears lots of academic departments begin September with students frantically searching on their subjects.

We can try another term with some currency:  “business analytics”.  The result is somewhat surprising.  “Thank you email”?  “Vendor portal”?  “Zoes Kitchen”?  It seems hard to make much sense of this.  As we know, “business analytics” is a relatively new term and the search quantity is less than that of “operations research” which perhaps explains the spurious correlations:  there are so many terms that are searched as often as “business analytics” that the highest correlations come more or less randomly.

To data people like us (me, anyway), the ability to search correlations is endlessly fascinating.  Shift the operations research time series by 13 weeks and what do you get:  things like “portable mp3” and “retriever pictures”:  clearly our students are bored with our course and are surfing around for something more entertaining.  What does “management science” search correlate with?  “introduction” and “social research”.  Is there anything interesting to be learned by the differences in correlates between operations research and management science?  Nothing springs to mind, but there might be a thesis or two there.

I am not sure what any of this means, but it sure is a great way to spend an early summer afternoon!

 

 

Business Analytics and Operations Research: Tomato, To-mah-toe, Tractor!

There are few things in life more tedious than assigning boundaries to fundamentally ill-defined concepts.  Either terms are used to divide things that cannot be divided (“No, no, that is reddish-purple and clearly not purplish-red!”) or are used to combine groups while ignoring any differences (Republicans?  Democrats?  just “Washington insiders”).  Arguing over the terms is fundamentally unsatisfying:  it rarely affects the underlying phenomena.

So when INFORMS (Institute of Operations Research and the Management Sciences), an organization of which I am proud to have been President and equally proud to be Fellow, embarks on its periodic nomenclature debate, ennui overwhelms.  Not again!  The initial debate between Operations Research and Management Science resulted in two societies (ORSA and TIMS) for forty years before they combined to form INFORMS in 1995.  Decision Engineering, Management Engineering, Operations Engineering, Management Decision Making, Information Engineering, and countless other terms have been proposed at times, and some have even made toeholds in the form of academic department names or other usages.  None of this has fundamentally changed our field, except perhaps in confusing possible collaborators and scaring off prospective members (“Wow, if they don’t even know who they are then maybe I should check out a more with-it field!”). I decided long ago to just stick with “operations research” and make faces of disgust whenever anyone wanted to engage the issue of the name of the field.

Then, three years ago (only! check the google trends graph) the phrase “business analytics” came along, and it was a miracle!  Here was the phrase that really described what we were doing:  using past data to predict the future and make better business decisions based on those predictions.  That’s us!  And, due to books such as “Competing on Analytics”, the wider world actually were interested in us!  There were even popular books like “The Numerati” about us.  We were finally popular!

Except it wasn’t really about “us” in operations research.  We are part of the business analytics story, but we are not the whole story, and I don’t think we are a particularly big part of the story.  A tremendous amount of what goes by the name “business analytics” are things like dashboards, business rules, text mining, predictive analytics,OLAP, and lots of other things that many “operations research” people don’t see as part of the field.  IBM’s Watson is a great analytics story, but it is not fundamentally an operations research story.  People in these areas of business analytics don’t see themselves as doing operations research.  Many of them don’t even identify with business analytics but rather with data mining, business intelligence, or other labels.   All of this involves “using past data to help predict the future to make better decisions” but “operations research” doesn’t own that aspect of the world.  There are lots of people out there who see this as their mandate but haven’t even heard of operations research, and really don’t care about that field.

This is not surprising for those with an INFORMS-centric point of view.  INFORMS does not (and near as I can tell, ever has) represent even all of “operations research”.  According to the Bureau of Labor Statistics, there are more than 65,000 people with the job “operations research analyst”.  INFORMS membership of a bit more than 10,000  is a small fraction of all those involved in operations research.  INFORMS is not all of operations research:  it certainly is a small amount of business analytics.  How can INFORMS “own” business analytics when it doesn’t even own operations research?

Recognizing this divide does not mean erecting a wall between the areas (see the first paragraph on the mendacity of labels).  I think the “business analytics” world has a tremendous amount to learn from the “operations research” world and vice versa.  Here are few things the two groups should know (and are clearly known by some on both sides, though not to an ideal extent);  I welcome your additions to these lists:

What Business Analytics People should Learn from Operations Research

  1. Getting data and “understanding” it is not enough.
  2. Predicting the future does not imply making better decisions.
  3. Lots of decisions are interlinked in complicated ways.  Simple rules are often not enough to reconcile those linkages.
  4. Handling risk is more than knowing about the risk or even modeling the risk.  See “stochastic optimization”.
  5. Organizations have been competing on and changed by analytics for a long, long time now.   See the Edelman competition to start.
  6. Operations research is not exactly an obscure field.  Check the google trends of “operations research” versus “business analytics” (with OR in blue and BA in red).

What Operations Research People should Learn from Business Analytics

  1. It is not just the volume of data that is important:  it is the velocity.  There is new data every day/hour/minute/second, making the traditional OR approach of “get data, model, implement” hopelessly old-fashioned.  Adapting in a sophisticated way to changing data is part of the implementation.
  2. Not everything is complicated.  Sometimes just getting great data and doing predictions followed by a simple decision model is enough to make better decisions.  Not everything requires an integer program, let alone a stochastic mixed integer nonlinear optimization.
  3. Models of data can involve more than means and variances, and even more than regression.
  4. One project that really changes a company is worth a dozen papers (or perhaps 100) in the professional literature.
  5. It is worthwhile for people to write about what is done in a way that real people can read it.

I believe strongly in both operations research and business analytics.  I have spent my career advancing “operations research” and have never shied from that name.  And I just led an effort to start an MBA-level track in business analytics track at the Tepper School.  This track includes operations research courses, but includes much more, including courses in data mining, probabilistic marketing models, information systems, and much more.

The lines between operations research and business analytics are undoubtedly blurred and further blurring is an admirable goal.  The more the two worlds understand each other, the more we can learn from each other.  INFORMS plays a tremendously important role in helping to blur the boundaries both by sharing the successes of the “operations research world” with the “business analytics” world, and by providing a conduit for information going the other way.  And this, more than “owning” business analytics, is what INFORMS and its members should be doing.

Ironically, this is part of the INFORMS Blog Challenge.

 

 

IBM, Ralph Gomory and Business Analytics

Had a post at the INFORMS Conference site on Ralph Gomory:

For those of us taking a break from the INFORMS conference, the Master’s golf tournament holds special attention. Not for the golf (though the golf is wonderful), but for the commercials. Practically every commercial break has an IBM commercial featuring some of its luminaries from the past. Prominent among them is Ralph Gomory. Everyone in operations research knows of Ralph. For the optimization-oriented types, he is the Gomory of Gomory cuts, a fundamental structure in integer programming. For the application-oriented types, he was the long-time head of research for IBM. For the funding and policy-oriented types, he was the long-time head of the Alfred P. Sloan Foundation supporting analysis on globalization, technology, and education. Great career, when you can be highly influential three different ways (so far)!

During his time at IBM, Ralph stressed the need for research and development to work together. This view that research should be grounded in real business needs is one that I think has greatly strengthened areas such as operations research and business analytics. While there is no dearth of theoretical underpinnings in these areas, the fundamental research is better by being guided by practical need. This has led to the insights that give us fast optimization codes, stronger approaches to risk and uncertainty, and the ability to handle huge amounts of data.

There is a full version of the IBM video that lasts about 30 minutes (currently on the front of their Smarter Planet page). Ralph shows up in the introduction, then around 24:43 in an extended discussion of the relationship between research and business need, and again near the end (30:08).

This conference would have been a lot different (and less interesting) without the career of Ralph as a researcher, executive and foundation leader. We are lucky he began in operations research.

Getting ready for INFORMS Business Analytics and OR conference

I’m getting ready for next week’s INFORMS Conference on Business Analytics and Operations Research. Looks like the renaming (from the INFORMS Practice Conference) has had an effect: the conference has gotten record registration (more than 600).

Getting ready for a conference is not just tossing some clothes in a suitcase. Keeping up my social networking responsibilities is a lot of work! I’ve changed my blog page to highlight the feed from the INFORMS Conference blog (where I will guest blog for a few days). We’ve started a discussion on the appropriate twitter-tag (I like #baor11). I’ve contacted some friends for suggestions of a brewpub to visit (Goose Island on Clybourn seems to be a good choice). Above all that, I have to read (thoroughly!) the papers associated with the Edelman competition, where I am a judge.

I have done my first post for the INFORMS Blog. Here is what I wrote:

I fly out to the Analytics conference in a few days. By some weird happenstance, I have never flown with Southwest before, but I am doing so on Saturday. In view of the issues Southwest is having, I need to do a bit of risk analysis. I really wish I could attend the risk analysis track before I get on the plane, instead of after I arrive.

Fortunately, Arnie Barnett (operations research go-to guy for aviation risk analysis) has provided insight into the risks. I think I’ll be OK with Southwest.

Great Way to Get to The INFORMS Conference on Business Analytics and Operations Research

I am very much looking forward to attending this year’s INFORMS Conference on Business Analytics and Operations Research (formally the INFORMS Practice Conference).  I am a judge for the Edelmans, so I will be spending Monday watching the presentations and asking tough questions (“Wow, did you really save $200 million?  That’s so cool!”).  I’ll also be attending some of the Technology Workshops on Sunday, and will attend other presentations on Tuesday.

Over the last few years, I have scrabbled together some funds to support sending some of the Tepper MBA students to the conference (thanks Tepper Administration for all of your support!), and they always come back raving about the conference and the field.  I expect this year to be no different:  we’ll have four students in our business analytics track (at least!) at the conference.

Two of the students will be attending the Professional Colloquium, a day-long program for Masters and PhD students who are transitioning into real-world careers.   I always worry when I suggest this to MBAs since the professional skills and insight into organizations that the day provides are the same skills an MBA provides (and which are more commonly lacking in normal masters programs in operations research).  Will they get enough out of the day? But every MBA who has attended the Colloquium has loved it:  the speakers provide insights into success from the perspective of operations research/business analytics professionals.  For many of the students who have attended, this is a life-changing experience.  I see that one of my students from a couple of years ago thinks enough of this to be part of this year’s organizing committee!

Whether you are a business analytics-oriented MBA, a Masters of OR or IE, or a Doctoral student (or a recent graduate in any of these areas), I can’t recommend the program highly enough.  And, while the registration fee of $375 might not seem cheap, it really is a steal, since it includes participation in the full conference as well as the Colloquium.  There are some limited support funds from the Colloquium committee, but this is the sort of activity that your school really should be supporting (and even if not, this is a great investment in your career).

Applications are due March 25, so get going if you want to be part of this!

New Year’s Resolutions from Dr. O.R. Field

In a tremendous coup for the blog, I am delighted to present this interview with Dr. Operations Research Field where she sums up the year 2010 and presents her resolutions for the upcoming year.

MT: Dr. Field …

OR: Please, call me O.R.: all my friends do, though some call me M.S. for some reason.  I’ve actually had lots of names in the past.

MT: OK, O.R., thanks for seeing me. I can’t help but notice that you are …

OR: A woman? Yes, well, as an artificial personification of an abstract idea, I can appear as I choose. This year, with the INFORMS President, President-Elect, five Vice Presidents, and interim Executive Director all women, it seems most representative of the field. Not to mention women like MIT’s interim Dean of Engineering, IBM Research’s Vice President of Business Analytics and Mathematical Sciences, and my three of my favorite bloggers (Anna, Laura and Aurelie). And we all know that WORMS has the best panel discussions and receptions!

MT: Well, you are looking very healthy. There were rumors that you were dying. How do you feel?

OR: Dying? Me? Not a chance! I’m sixty years old and healthy as a horse. People have been talking about me dying for decades, and I keep outliving them all. I don’t think there is any chance of me dying anytime soon.

MT: Funny, I could have sworn you were dying. Any proof that you are actually healthy?

OR: Look at the parties I throw every year (that is “professional meetings” when discussing reimbursements with department heads). The last party I threw in Austin drew more than 4600 people! I remember ten years ago I would hold two parties a year and be lucky to get 2000 at either one of them. Any my European parties (and what parties they are!) are also breaking records every year.

MT: Wow, that is pretty amazing. Any other signs of success?

OR: Well, the Edelman Awards showed how international I got, with finalists from the U.S., Mexico, Germany, Canada, and South Africa. INDEVAL from Mexico won with a nifty way of optimally clearing trades in a financial market, meaning financial firms didn’t have to leave loads of cash lying around.

MT: That was a cool paper.  What else?

OR: I love to see all the blogging, tweeting, and other things going on.  OR-Exchange looks to be getting a critical mass of participants, replacing the dear departed sci.op-research.

MT: Hey, that’s not dead!

OR: It does seem to be the place to go if you want a solution manual or cheap shoes.  But lots of other things are replacing it, so I think I’ll survive its loss.

MT Well, what is up for Dr. O.R. Field in the upcoming year? Have you made any New Year’s resolutions?

OR: Hey, that’s pretty clever. By bringing in the holiday, you can enter this post in INFORMS’ December Blog Challenge!

As it happens, I have made resolutions for the upcoming year. Here are a few of them

First, I resolve to embrace and define business analytics.  But, as I recently read…

MT: If people write me, I’ll let you know where you read this..

OR: …Thanks… there is a tremendous risk that business analytics will be conflated with predictive analytics, like data mining, just as the line between business intelligence and descriptive analytics became very fuzzy.  Business analytics has got to go beyond prediction to actual decision making, prescriptive analytics if you will, where complicated decisions are made based on the results of predictive and descriptive analytics.  So the best business analytics will go beyond “Should I send this person a catalog” to “Based on these three predictive models, how should I optimally allocated my marketing budget in the face of these various constraints”.  It is that level that I have the most to add.  And I’d really like to be more famous!

MT: I agree!  I wrote about that in the context of revenue management:  you can use data mining models to provide real-time input into a hotel reservation system to provide a much more profitable and effective overbooking system.  You really need to take ownership of this!

OR: I hope that the revamped INFORMS conference in the spring will really have an effect on how people think about me and business analytics.

My second resolution is to pay more attention to robustness in my models.

MT: You mean like the work of Bertsimas, Ben-Tal and many others?

OR: That is one type of robustness, but I meant something a bit broader.  Let me give you an example:  I was recently flying to one of my parties and had to change planes in Detroit.  Now, a bunch of things have to come together in order for the flight to go.  You need a plane, a pilot, a first officer, a flight attendant and so on on.  Now, if the pilot is on a flight coming in from Pittsburgh, the first officer is on a flight from Montreal, the flight attendant is coming in from Des Moines, and the plane is from Albuquerque, then what is the chance that my flight is going to go?

MT: Pretty small?

OR: Right! Zilch, zero, nix, nada, zippo.  There is no way that plane is going out on time, since it is delayed if even one of those multiple planes is late.  This system may be optimized, but it is certainly not robust.

But if the plane and the whole crew worked together, then my flight would have been late only if one previous plane was late.  This is a much more robust system.

MT: I blogged about David Ryan’s work on this years ago.

OR: Well, I guess no one reads your blog, because it is a huge issue.  Imagine if the financial system was designed with robustness in mind.  We might not know where the uncertainty is coming from (just like we don’t know what plane is late), but the system would better adapt to whatever weird things happen.

MT: Kinda like methods for avoiding bus bunching.

OR: You bet, but on a much larger scale.

My third resolution is to do a better job at getting the word out about me and the wonderful things I do.  We’ve got a few bloggers and a magazine or two but we have to do more!

MT: Well, I’ll do my part!

OR: I hope so.  I don’t want to see any  more disgraces like this year when you had only one blog post in August and two in October.  I almost unfriended you on Facebook for that!

MT: Sorry, I guess I better put that down as my resolution.

So, O.R. any final comments?

OR: Well, three resolutions don’t seem to be enough for me:  I’d welcome any resolutions people might suggest.  And may you all have an optimal New Year!

MT: Happy New Year from me too!  And feel free to add your resolutions to the comments and I will be sure that O.R. sees them.

The End of the INFORMS Practice Conference …

and the start of the INFORMS Conference on Business Analytics and Operations Research.

The INFORMS Practice Conference has long been one of my favorite conferences.  In addition to the inspirational Edelman Competition presentations, the organizers do a great job of identifying presenters for a range of industries, illustrating the wide applicability of operations research.  The conference is on a much more manageable scale than the INFORMS Annual Meeting (or EURO meetings) and is typically held in an interesting location.

The INFORMS blog  has just announced that the name for this conference series will change to the INFORMS Conference on Business Analytics and Operations Research.  Business analytics is a term that has gained a lot of recognition recently (we recently added a track to our MBA program called “business analytics”).  INFORMS has a pretty good definition of the term:

Business analytics facilitates realization of business objectives through reporting of data to analyze trends, creating predictive models for forecasting and optimizing business processes for enhanced performance.

While all of those aspects are “operations research” it is that last phrase “optimizing business processes” that really links business analytics to the OR/MS world. Previous approaches like “business intelligence” did not really integrate aspects of optimizing business processes.  But with this definition “business analytics” really is “operations research” and vice-versa.

I have struggled with the adoption of the phrase “business analytics”.  INFORMS (and its founding organizations ORSA and TIMS) has had innumerable discussions on what our field should be named and even now INFORMS embeds two alternatives: Operations Research and Management Science, the ORMS of INFORMS.  Do we need another name?  And there are aspects of “business analytics” that seem to me to be a stretch to call operations research:  once you start tossing in dashboards and scorecards and the rest of the buzzwords, I start racing back to my safe world of cutting planes and submodular functions.  But it is all about using data and models to make better decisions.  And if the market likes “business analytics” then I’m good with it.

I worry about the longevity of the term.  Will this term last or in five years will it feel like “e-business” does today?  It does strike me as a term that has the chance of being around for a while, particularly if it is embraced by organizations like INFORMS.

One big problem for the phrase:  there is no good phrase to identify those that do it.  “Business analysts” is not right:  analyst comes from analysis, not analytics.  “Business analytickers?”  I think not.  On the other hand, “operations research” has that problem too.  “Operations researchers” just doesn’t sound right.

So I am good with the title.  However, can we call it the “INFORMS Conference on Business Analytics and Operations Research”, its official name, not the “INFORMS Analytics Conference” as given in the INFORMS blog title?  Including the “operations research” name is going to be important to the branding of this conference.  At least as far as us ORers are concerned.