Software // Information Management
Commentary
10/26/2009
09:29 AM
Neil Raden
Neil Raden
Commentary
Connect Directly
RSS
E-Mail
50%
50%

Who Needs Analytics PhDs? Grow Your Own

Analytics, whatever that means, has emerged as the hot topic all over our industry... According to the conventional wisdom, very special experts, quants we'll call them, are needed because mere mortals can't handle this stuff... But I don't buy this.

Analytics, whatever that means, has emerged as the hot topic all over our industry. Gartner seems to have bolted from Business Intelligence and placed "Advanced Analytics" in the firmament of must-have technologies for 2010 (I guess everyone followed their lead and implemented BI last year so there is nothing else to talk about). The problem with analytics is, who can do it? Numerate people in organizations are as scarce as hen's teeth. According to the conventional wisdom, very special experts, quants we'll call them, are needed because mere mortals can't handle this stuff. But you can't buy quants like muskmelons on the road to Bettendorf in July, and even if you could, a muskmelon would be less troublesome.One of the often-cited problems with quants is that they tend to be condescending towards those who lack quantitative skills. This is the reason Tom Davenport often uses the term "PhDs with personalities," a phrase I find quite distasteful. However, most people's experience with quants leaves them yearning for someone who speaks in a common language and doesn't get impatient when asked stupid questions. But I don't buy this. Quants don't act any differently than other highly trained professionals who have to repeatedly explain themselves to those who have no background in the subject and show no initiative in acquiring it. IT people for instance. Sound familiar? In addition, why all this talk about PhDs anyway? There is nothing so special about most "advanced analytics" that someone with adequate training could not do. Actuaries are a good example. Most actuaries can (and do) apply very complicated quantitative methods to their work, yet very few have a PhD. A bachelors or masters degree in math, physics, biology -- anything that took them through, say, four semesters of calculus, plus linear algebra, differential equations and real analysis. These subjects are table stakes for almost any career in math, engineering or the sciences. I didn't mention probability and statistics, because I was an actuary and did not take those courses in college. I, like many actuaries, learned those subjects and passed the exams through self-study over a number of years. Insurance companies give actuaries study time (one to two hours/day) for the three to five years it takes to pass all the exams. Why not do the same for quants? The actuarial exams are pretty brutal, but there is no reason for a quantitative expert course of study to emulate that. I'd urge companies to set up such a plan with an eye to ripening experts over a two- to three-year period with time at work to study. Hire promising people out of college with the right background, and allow them to grow into the role of a rocket scientist super quant. One more thing: When you get a PhD in math or statistics, you get a research degree, not a clinical practice degree. That means, in order to get it, you have to spend some years on original research and defend it. The keyword is original. By the time someone gets a PhD, their field of interest is so narrow, they probably have no better grasp of the types of analytics that a commercial company needs than a BA/BS math major with adequate training. The real mental challenge is in developing new algorithms and models, which is where PhDs belong. Building a pricing model using multiple regression or Bayesian analysis is not. This PhD stumbling block is a result of collective fear of math -- it's hard so let's get the smartest people on it. But the smartest people may be standing right in front you.Analytics, whatever that means, has emerged as the hot topic all over our industry... According to the conventional wisdom, very special experts, quants we'll call them, are needed because mere mortals can't handle this stuff... But I don't buy this.

Comment  | 
Print  | 
More Insights
Comments
Newest First  |  Oldest First  |  Threaded View
neil raden
50%
50%
neil raden,
User Rank: Apprentice
3/31/2012 | 8:20:12 PM
re: Who Needs Analytics PhDs? Grow Your Own
Doug,

For any reader not familiar with the abbreviation QED, it is not the call letters of a Pittsburgh public television station. It stands for the Latin phrase for "quod erat demonstrandum," and is typically placed at the end of a mathematical proof or philosophical argument. So it is with some hesitation that I respond since Doug indicated the discussion is over.

Job descriptions represent the current miscomprehension of data scientist. The term is a another example of title inflation, along the lines of "Chief." At Google, with a PhD in Noncommutative Markov semigroups from MIT, and five years experience supporting portfolio managers on Wall Street - I can abide that. Someone who know SAS and has been building propensity models for a B2B site, not. My job isn't to catalog the current miscomprehension, it's to inform it.

-NR
Doug Laney
50%
50%
Doug Laney,
User Rank: Apprentice
12/7/2011 | 10:57:34 PM
re: Who Needs Analytics PhDs? Grow Your Own
To clarify, Gartner's position is, while we don't split hairs and there's accepted overlap and personal vernacular preferences, that we use "BI" more to define tools and platforms, "analytics" in referring to the overall discipline, and "advanced analytics" in reference to really anything beyond pedestrian query & reporting, i.e. stuff mere mortals most certainly cannot do. As for the role of the data scientist, anyone who can model business problems, gather and prep data, develop and test hypothesis, drive SAS, SPSS, R, etc. and persuasively articulate findings can feel free to call him/herself a data scientist.

I guess we'll let reality speak for itself. Here's a quickndirty wordcloud analysis I did comparing dozens of "data scientist" job descriptions with "business intelligence analyst" job descriptions: https://docs.google.com/presen.... Hiring orgs are certainly distinguishing the two.

Q.E.D.

-Doug Laney, VP Research, Gartner
The Agile Archive
The Agile Archive
When it comes to managing data, donít look at backup and archiving systems as burdens and cost centers. A well-designed archive can enhance data protection and restores, ease search and e-discovery efforts, and save money by intelligently moving data from expensive primary storage systems.
Register for InformationWeek Newsletters
White Papers
Current Issue
InformationWeek Tech Digest - September 10, 2014
A high-scale relational database? NoSQL database? Hadoop? Event-processing technology? When it comes to big data, one size doesn't fit all. Here's how to decide.
Flash Poll
Video
Slideshows
Twitter Feed
InformationWeek Radio
Sponsored Live Streaming Video
Everything You've Been Told About Mobility Is Wrong
Attend this video symposium with Sean Wisdom, Global Director of Mobility Solutions, and learn about how you can harness powerful new products to mobilize your business potential.