Data science has been one of the fastest-growing employment fields, with demand for data scientists growing by 650% from 2012 to 2017 alone, according to LinkedIn.
Demand is expected to remain that way; from 2020 to 2030, the job outlook within the field is projected to grow 22%, according to the U.S. Bureau of Labor Statistics.
To address this demand, colleges and universities have launched new master’s degree programs in data science. In fact, Howard University, the University of Connecticut, and the University of California San Diego have all announced new programs in recent months.
With so much changing so quickly, what does the future hold? Fortune spoke with data scientists to learn how data science programs will likely evolve.
Staying up-to-date with technology
The field has transformed even in the nearly seven years since Grant Long completed his master’s degree in data science from New York University.
“It’s changed tremendously,” says Long, now the head of data and machine learning at Flowcode, a startup that aims to bridge the offline and online worlds through QR codes, mobile-focused landing pages, and other digital tools.
“Technology continues to change. I think a lot of data science resources are organized around open-source packages which have become more and more popular across a variety of companies. Obviously [at] tech companies where a lot of these open-source projects originate, but also [at] more traditional companies.”
Long says the adoption of this new technology has been “transformational” for many businesses. While programming models like MapReduce were taught when Long was at NYU, it’s more likely that master’s degree students are now instructed on open-source programs like Apache Spark.
Although technology may change, the fundamentals of data science remain the same, says Joel Sokol, director of the interdisciplinary master of science in analytics degree at Georgia Tech and a professor in the university’s H. Milton Stewart School of Industrial and Systems Engineering.
“In some sense, there’s not that much distance from where you start to where you’re already at the cutting edge,” such as natural language processing and deep learning, Sokol says.
“The biggest thing that schools are trying to do in these programs is try to keep up with what’s at the cutting edge because that’s where a lot of professional practice ends up.”
With so much new development in the field, Sokol says it’s common to see the faculty of other schools taking online courses from Georgia Tech, Carnegie Mellon University, and Massachusetts Institute of Technology to stay on top of the latest advances.
As for how artificial intelligence is impacting the field of data science, Sokol cautions that what some people call artificial intelligence today is essentially stronger machine learning algorithms, not something like HAL 9000 from “2001: A Space Odyssey.” Still, he says, these algorithms will open some new doors for data science, potentially assisting the ability to make predictions.
How data science programs focus on practical versus theoretical
Within the field of data science, there’s already been a bit of a bifurcation between practitioners who focus on the practical side and those who focus on the theoretical side. Sokol says that as there aren’t many Ph.D. programs in data science, there will continue to be more specialization at the graduate level.
“You’ll end up with some master’s programs that are designed for practitioners, where they focus on the application of these methods and how to use them, and then some master’s programs are going to be essentially pre-Ph.D. programs, where you learn more about the theory,” Sokol says.
Additionally, Sokol says there are data science and analytics certification options. In these non-master’s programs, participants learn the basics of programs and software but little about the theories and approaches behind them.
These “boot camp” options will continue to see increased demand for a while as there aren’t enough people in master’s programs to fill the current need for data science and analytics positions; however, Sokol says that as more people go through graduate and undergrad programs that prepare them for these roles, these boot camps will see less demand.
With a career that includes stints at the Federal Reserve Bank of New York, Capital One, and Zillow, Long says he’s witnessed the increase in specialization first-hand. While at Capital One in the mid-2010s, Long says data science was considered a broader term, and that the role of a data engineer was just starting to emerge. At the time, data scientists were doing a lot of data engineering work.
Now, at Long’s startup, there are five different titles among the roughly 85 full-time employees in the field of data science: machine learning engineers, analytics engineers, data analysts, product analysts, and data engineers.
Long says data science programs of the future will further specialize into computer science and statistics at a rigorous technical level, as well as analysis. With the growth of packaged data science solutions that require little coding, like Alteryx and AutoML, non-data scientists can do simple analyses of data, but Long says there will continue to be a high demand for people who can conduct more complex ways of capturing and analyzing data. That need will continue to be seen in data science graduate programs.
“Going forward, there’s going to be greater specialization of the program,” Long says. “There will also continue to be stronger [rewards] to programs that emphasize mathematical and computer science fundamentals because those things are not easy to pick up.”
Because technology is changing so quickly, employers are looking for people with a stronger background in manipulating data and are rewarding them with higher rates of pay. In fact, some students who are graduating this spring are already fielding job offers of $125,000 and up. Programs that emphasize this foundational teaching will see graduates who are paid more obtain higher profile jobs and see greater demand from program applicants, Long says.
An emphasis on real-world experience
A master’s degree in data science helps signal that a job candidate knows the essentials, but Long says that potential hires are more impressive when they combine that degree with real-world experience.
“What I look for when I see candidates is that they have some exposure to data in some role before, but have also had some exposure to fundamental theories and computer science and math,” Long says. “That helps signal that you’ve had that fundamental exposure.”
While open-source packages have made some elements of being a data scientist easier, Long says it’s still critical to know the basics.
“At a place like Flowcode, the most important thing is to have people that are trained to be adaptable, which is why some of the most classic elements of computer science and math are actually the most important,” Long says.
