[Image of: RetailWire Logo and Tagline (for print)]

Will the need for data scientists change America's educational model?

April 23, 2014

In a discussion last month on RetailWire, Nikki Baird of RSR Research wrote that finding a data scientist was akin to finding unicorns and dragons in retail enterprises. There is no doubt that the explosion of big data has created a real need for numbers gurus who can make sense of key points and communicate insights to the rest of the business organization.

So what skills does it take to become a data scientist and where do companies find the people that have them?

According to a recent article on the Pacific Standard website, companies have gone in search of individuals holding advance degrees in the past. Today, however, becoming a data scientist requires basic math skills and a computer connected to the internet. "Becoming a data scientist," according to the article, "is perhaps the most prominent example of a new industry that breaks from the higher education model and allows people to learn the necessary skills without years of classes."

The Pacific Standard article profiles three data scientists who took statistics in school, but otherwise are largely self-taught. Brian Burke, a former Navy fight pilot and military contractor, founded Advanced NFL Stats. Charles Pensig is a senior data analyst at Jawbone, the wearable tech and audio device company. Carl Bialik, is lead news writer at Nate Silver's FiveThirtyEight. None of the three suggested that the higher educational model is the means to becoming a data scientist.

"There is a treasure trove of information available on the Web, most of which is far more gentle, user-friendly, and effective than a grad school course," Mr. Burke told Pacific Standard. "Free courses on Coursera or similar sites can be really great sources."

Discussion Questions:

Do you agree that people can be taught how to be data scientists without years spent pursuing advanced degrees? Will the need for data scientists help to spur changes in America's colleges and universities?

While we value unfettered opinion, we urge you to show respect and courtesy for people or companies about whom you comment. Keep in mind that this is a public, professional business discussion. RetailWire reserves the right to edit or refuse the publication of remarks that we deem unsuitable. We may also correct for unintended spelling and grammatical errors.

Instant Poll:

Do you agree or disagree that data scientists do not require extensive formal education to succeed in their jobs?


I hear this same challenge when I attend events and speak with retailers. At a recent event, an exec from Gilt Groupe mentioned their talented data scientists and I asked if he could let other retailers know where to find good data scientists — and he basically said "No."

I do think colleges and universities need to step it up when it comes to this expertise. It's a great opportunity for young people looking for a potentially lucrative career.

[Image of: View Braintrust Panelist button]
Debbie Hauss, Editor-in-Chief, Retail TouchPoints

Yes, you can learn what you need without pursuing an advanced degree, but you still need the knowledge you would get from an advanced degree. I'm guessing your knowledge search would be more focused — you could avoid the theoretical stat courses I had to take. Just learning stats, however, does not a data scientist make — business knowledge and presentation skills are just as important.

[Image of: View Braintrust Panelist button]
Dr. Stephen Needel, Managing Partner, Advanced Simulations

The challenge and value of true data scientists are their ability to use complex concepts and mathematics to uncover trends and significant insights that translate to predictive recommendations from vast amounts of data points. These concepts and their derivations are not something you learn online. Data scientists are not technicians or statisticians.

I am fortunate to have a true data scientist on our StoreStream Metrics team who had a significant role in designing and implementing the recommendation engine used by Amazon.

All too often I find folks who claim "they have the people" to do this work yet I don't see retailers leveraging this degree of complexity. They believe their 'research' department is "doing this." The resources do exist for amazing results to occur if retailers would rise above the politics and status-quo to leverage this new paradigm.

As with anything else, the rules of supply and demand will dictate the fate of true data scientists. As data continues to exponentially grow so too will the demand for highly skilled people to uncover the insights this massive data pool holds.

[Image of: View Braintrust Panelist button]
Adrian Weidmann, Principal, StoreStream Metrics, LLC

I'm always looking for gifted programmers and systems engineers for a software venture and have actually been advised against hiring people with advanced degrees, especially PhDs.! That's one of those unfair judgements with just an element of truth.

Personally, I have a genetic aversion to anything even resembling a spread sheet. But just this morning I watched a Business Insider video on how to use an Excel trick to make practical sense of data. Very cool stuff indeed. Chances are a retail shop already has a very bright 20 something employee who could do data analysis in heart beat. Here's the link to the video.

[Image of: View Braintrust Panelist button]
Ian Percy, President, The Ian Percy Corporation

Those years spent pursuing advanced degrees do more than just give the scientists an education. They provide networking opportunities, chances to succeed (and fail) on many projects and perhaps most important experience with technology solutions. While some individuals may not need the structure or even the curriculum of an advanced degree, namely stat prodigies and math geeks, there really isn't a substitute for education.

On the second question, these changes are already happening. The notoriety and success of stat/data whizzes like Nate Silver, as well as forces in the marketplace, has pushed academia to offer new courses and even full degrees in data and predictive analytics.

[Image of: View Braintrust Panelist button]
Ron Margulis, Managing Director, RAM Communications

I can only draw conclusions here from personal experience: My son is a data analyst by trade, not because he intended to specialize in the field, but because his first employer (where he began as a software developer five years ago) recognized his aptitude. It helped, of course, that he was a math and computer science major who took a statistics class, so he -- like many others in the field -- would naturally be drawn to data analytics.

It's a small sample size, but companies needing to build their data analysis teams (with growing management of "big data" to predict behavior and outcomes) should identify those with an innate interest and skill set, not just those who have sought out advanced degrees. At the same time, the field of data analytics is growing so fast that colleges with "applied math" specialties ought to provide more training for undergrads heading into the workforce.

[Image of: View Braintrust Panelist button]
Dick Seesel, Principal, Retailing In Focus LLC

Retailers probably just aren't looking in the right places for people with these skill sets. They are definitely out there -- a friend of mine has a Ph.D. in cognitive psych and works as a data scientist for a large national retailer. I don't know if there is really a need for specific data science programs when it is already a major component of existing academic and scientific disciplines. The talent is out there; companies just need to get more imaginative and flexible with their recruiting (not to mention being willing to offer competitive compensation).


Yes. However, we shouldn't discount the advantage that an advanced degree offers, including learning from the mistakes and experiences of others (often first-hand), being taught differing perspectives on the same process or procedure, discovering new perspectives which are not widely available, etc. Combine all of this with peripheral courses which may compliment analyzing data, and you have the basis for an advanced degreed platform at a university.

[Image of: View Braintrust Panelist button]
Kai Clarke, CEO, American Retail Consultants

We (colleges and universities) must do a better job preparing students for the big data/analytics world. Every business discipline and every company must deal with a growing volume, velocity, and variety of data.

In response to this need, we started requiring 6 hours of business analytics for all business students (at Auburn University's Harbert College of Business) a couple of years ago. Starting last year, we now offer an undergraduate major in business analytics, a business minor in BA, and an MBA concentration in BA. Should the need for data scientists spur changes in America's colleges and universities? We certainly believe it should.

Bill Hardgrave, Dean, Auburn University

As someone who built a data science team, I know the challenges of finding good people. The problem with universities is that a lot of them do not know what makes a good data scientist for industry. So they create programs they want to teach, not really what we want to hire.

I have found that in a lot of cases, the skills do exist to some degree with people already in the workforce. What we did was train up analysts in the areas they were short in. We didn't need to send them in for a degree, just a few classes to improve their skills. It saved us time and money.

Also, the role of a data scientist 4 years from now will probably be very different than today. So anyone going to school to get a degree may find that the world changed on them while they were in school.

Edward Chenard, Innovation Lab Leader, Target

Changes in universities are happening. Every college of business is having this conversation. Many programs are being developed, tested, and implemented. In full disclosure I will tell you that I am working on such programs at my university and am working to develop tools as part of the Executive Committee of the Teradata University Network (TUN) Advisory Board.

There are a number of issues: 1) many faculty, especially in the application disciplines such as marketing, have not been trained in data analytics, 2) there are very few tools available to use for giving students experience. In my work with faculty around the world and as part of TUN, the biggest problem is lack of data sets to use for classroom instruction. Many of the company representatives at meetings about this issue have volunteered their company's data only to find that when they request permission from management, the request is denied. Without data to use when working with the tools and software, the training is not as effective as it should be.

There are many definitions and many different job descriptions for data scientists. People can perform some of the jobs without an advanced degree if they have specialized training. Many of the jobs require the experience, maturity, and critical thinking skills that come with an advanced degree. Some jobs require those skills plus specific advanced training in statistics, modeling, and/or computer science.

[Image of: View Braintrust Panelist button]
Camille P. Schuster, Ph.D., President, Global Collaborations, Inc.

Data scientists can be taught, just as can dragons and unicorns, with years spent pursuing advanced degrees. However, time spent in college has many other advantages as pointed out by Ron above.

As - or if - the need for more data scientists arises, colleges, retailers and industry experts will all spur changes in their specific environments. Like so many examples of accomplishments in history, this challenge too shall pass.

Gene Hoffman, President/CEO, Corporate Strategies International

Many of today's best universities are seeing data analysis as a North Star leading to new curriculum development. But in order to reach their goal successfully, they will need a moral compass.

This big trend is in direct correlation to the emergence of Big Data, and all the headlines that go with it. Last September I had written about the pending demand for data experts in the loyalty-marketing field. Among the figures cited: 97 percent of companies with revenue of more than $100 million are pursuing expertise in business analytics, according to Forrester Research. Yet the data analytics field is forecast to fall short of professionals by 2018, with an estimated 190,000 qualified data scientist positions left vacant at that time.

Colleges are scrambling to attract and educate students fast enough to meet this burgeoning demand. At North Carolina State, all 84 of last year's graduates were offered jobs, the program's director told the Times. The average salary exceeded $89,000.

So how do these universities build the models that balance these considerations while sorting out those who want to solely let the data speak for itself? A recent NY Times story answered it in one sentence: "Ethics classes address these questions."

It's almost a throwaway line, but it resonated. Not a single person should have access to data without understanding the rationale for having it, the need to care for it, and the implications of mismanaging it. There are tremendous opportunities for both companies and customers that arise from sharing information and improving the way we create relevant interactions. But the industry is simply expanding too fast to cut corners, and our brands, reputations and balance sheets are at risk if we don't ensure evenness between what is possible and what is right.

This new frontier of professional possibilities may have been brought to us by Big Data. But the inclusion of formalized ethics courses, designed specifically to address the considerations and possible repercussions of data use, is a good step forward. Combine that with an ongoing dialog around best practices, and we should have the ingredients to help us all make the move from Big Data toward good data.

[Image of: View Braintrust Panelist button]
Bryan Pearson, President and CEO, LoyaltyOne

Okay, I can answer this one. You see, I am one of those folks who just loves data! Had there been a job for data scientists back when I was in school, I would have been first in line. I did look at actuarial jobs back in the day and that was pretty close.

Does it require an advanced degree? Absolutely NOT! Excuse me for saying this but the fancy-pants data people with the big degrees are too textbook to be any real value for what we in retail need. There, the truth is out!

What it takes is someone with a real love and understanding of the nuances of data, but you don't stop there. The other critical piece is having the master insight person right beside that data scientist to bring wisdom to the process. The success of the team will lie in the give and take between those parties.

If you don't believe me, read Malcolm Gladwell's book "Blink."

And that's my 2 cents....

Lee Kent, Sharing Insights for Success in Retail, YourRetailAuthority

There' s truth in all the answers above. Yes, I believe that people can be data scientists without years spent pursuing advanced degrees. Yes, the need for data scientists will probably spur changes in America's colleges and universities. Formal education does not an expert make, but it will help.

As many of the respondents suggested, the best data scientist will have a combination of skills plus experience, plus interest, plus networking, plus objectivity, plus ability to communicate, etc. The coming generations who grew up with a computer in their hands (actually, starting with computers in their cribs, nay, computers in their mother's tummies before they were born) will have ample basic skills to start in the directions needed, tempered by a lot of experience and knowledge. Just having the capability won't count for much if it can't be applied and used practically.

[Image of: View Braintrust Panelist button]
Naomi K. Shapiro, Strategic Market Communications, Upstream Commerce

As recently pointed out in an article in Fortune, a great deal of the data being discussed here is "found data." The potential risks and liabilities of found data are far higher than data obtained from other sources. Found data can never be considered comprehensive and abounds with hidden skews.

So, for example, paleontologists when I was a kid confidently told us that dinosaurs were reptilian loners who moved slowly. Why? They were working with found data. When new data was found, it turned out that many dinosaurs were more like birds - meaning they were quick, vicious predators and some even appear to have had family groups or herds.

What this means is that data scientists need extensive training in the challenges of data collection as well as analysis. And a degree is the best place to get that training.

That said, a degree is also no guarantee of accuracy - since many research firms struggle with valid interpretation even under the best of circumstances.

And change the universities? Remember the 1980s when we needed to "re-invent education" to fill all those new jobs in tech? Well, re-inventing took a couple decades.... In the meantime, those jobs went off-shore for economic reasons or were replaced by automation. And, so, today the US has a glut of tech trainees working as baristas.

The risk of data analysis becoming an off-shore boom is quite high - and if it happens for economic reasons then increased graduates in the US won't change that. So let's be very cautious before dashing off after what looks likely to become an educational fad.

That said, as a trained mathematician (MA - UC, Applied Math) I know there is a vast existing training network already in place - we just need to use it better.

Doug Garnett, Founder & CEO, Atomic Direct

Search RetailWire
Follow Us...
[Image of:  Twitter Icon] [Image of:  Facebook Icon] [Image of:  LinkedIn Icon] [Image of:  RSS Icon]

Getting Started video!

View this quick tutorial and learn all the essentials...

RetailWire Newsletters