Two months of all-expenses-paid training. A six-figure income. Guaranteed placement. These are the temptations of several data-science boot camps that have sprung up over the last couple of years aiming to transition advanced academics to the private-sector title of "data scientist" that was still largely unknown in 2009, when Google economist Hal Varian characterized "statistician" as "the sexy job of the next 10 years".
Like statisticians, data scientists attempt to extract meaning from information. Where they differ, according to insideBIGDATA managing editor Daniel Gutierrez, are in tools used, variety of data sources exploited, and emphasis on business optimization. Where a statistician might create actuarial tables using SPSS and census data, for example, a data scientist is more likely to create a recommendation engine using machine-learning algorithms and live shopping data.
The enormous growth (and ready availability) of data has propelled demand for data scientists beyond what the market can provide, according to a 2011 report by management consulting firm McKinsey & Company that predicted "a shortage of 140,000 to 190,000 people with deep analytical skills" in the U.S. by 2018.
Rather than training employees in the complex field of data science, data-science boot camps teach business skills to math and science academics in their final year of Ph.D. study, or later. For some programs, a student pays little or nothing; sponsors -- who are also potential employers -- pick up the majority (or all) of the tab. A common boot camp structure involves a week or two of training in such subjects as tools (Hadoop, databases); computer-science languages and practices; machine learning; data visualization, and business standards. Students then form small teams and spend the bulk of their remaining time working on a real-world project provided by a course sponsor, who may use these efforts in its real-world business. Finally, students prepare for the professional world by packaging their experience and preparing for interviews.
Often the sponsoring companies hire members of the project teams outright. Gary Richardson, director of Data and Analytics Engineering at professional services firm KPMG (U.K.), explains why KPMG became principal partner of the Science to Data Science (S2DS) program for 2014, helping cover the costs of student training, technology, and accommodations in London: "We took the view that this was an innovative way to attract top data science talent into KPMG in the U.K.," he says. "Traditional recruitment agents struggle with the data science brief, take a large fee, and provide candidates who are an unknown quantity. Whereas here, we were able to assess potential hires over the five-week period."
Benefits gained from student projects comprised the main draw for predictive marketing startup Growth Intelligence, according to data team lead Alex Mitchell. "As a small company, we really wanted to get some concrete output from our project," he says of the company's S2DS sponsorship. "So we waited until the last possible minute to decide what to do, so that it was as relevant as possible to our customers. Then I put quite a lot of time into managing the project."
Students often need that hands-on guidance, according to Kim Nilsson, managing director of Pivigo Academy (which runs the S2DS program) and co-founder of Pivigo Recruitment. "It's always a culture shock," she says. "I think there are two main differences between academic and commercial environments. First, deadlines; a company might say, 'deliver this next week, and I don't care whether you think it's feasible or not.' In academia it tends to be more like, 'Work on it until you think it's done.'
"The second difference is in teamwork. In academia that means, 'I'm going to work at my computer, then we're going to meet for coffee for half an hour, then I'm going to go back to my computer.' In industry it's, 'we're going to sit together with our laptops, 100%'."
Does data science really require a new training approach? "Data Mining for Dummies" author Meta S. Brown has some doubts. "It sounds like these boot camps are good educations," she says, "But I do think it's a gold rush. On the one hand, there's a genuine need for people who know how to analyze data. But the proliferation of programs that say data science, analytics, whatever, reminds me very much of the MBA expansion in the 1980s. An MBA had been a relatively uncommon degree, but now it's hard to find any post-secondary school that doesn't have an MBA program, and there are an awful lot of MBAs who are not seeing the same rewards that drove the growth in those programs in the first place."
Meanwhile, continuing studies and extension courses in some traditional universities are experiencing enormous demand for their data-science courses. "Normally, we worry about how to fill classes when we launch a new program," says Salman Kureishy, program director of Business and Professional Studies at the University of Toronto School Of Continuing Studies. "For our recently launched certificate in Enterprise Data Analytics, the initial response has been overwhelming." He's not worried about losing students to the boot camps, as he sees substantial differences between their offerings and those of his school. "A university provides a more broad-based education and accepts students at all levels. Boot camps provide training, and assume you already have a certain level of prior knowledge."
For those who possess that level of knowledge, boot camps can represent a welcome exit from the uncertainties of academic life. "With money and austerity and everything else the way it is, research funds are really challenging to come by at the moment," says recent S2DS graduate Adam Hill, an astrophysicist currently working on the EU-funded Marie Skodowska-Curie actions Research Fellowship. "It's the risk of being an academic: you never know where the next pot of money is coming from, so having an exit strategy for going out into the real world, as it were, is always a good idea."
Tom Geller is an Oberlin, Ohio-based writer and documentary producer.
No entries found