Data science is a team sport, how do you choose the right players?

Building deep and enduring data science capabilities is not an easy process – it requires the right people, the right processes, and the right technology. Finding the right people for the right jobs has proven to be an ongoing challenge, as evidenced by employers and job seekers.

Several experts discussed this topic during a round table organized by Domino Data Lab. “The human side is probably the least understood aspect of this whole equation,” says John Thompson, global head of advanced analytics and artificial intelligence at CSL Behring.

As Head of Analytics at one of the world’s leading biotech companies, John Thompson oversees data science teams that deal with a wide range of initiatives. The virtual roundtable guests agreed that measuring data science requires more than data scientists.

Data science is practiced in teams

To launch data science initiatives at CSL Behring, John Thompson says it “starts with the backbone team you need to make a project successful.” This team generally consists of engineers, data scientists, data visualization for UI or UX, and subject matter experts.

For a data science team to be effective, it also needs a leader who can ensure projects remain focused on business goals.

“If we say that data science is a team sport, you don’t just need all the players, you need a coach,” agrees Matt Aslett, Director of Data Research, IA and Analytics at 451 Research.

Stay away from information technology

Obviously, the full data science team isn’t just made up of data scientists, but it doesn’t necessarily make sense to assemble data science teams in the IT department, adds Nick Elbrin, CEO and Co-Founder. by Domino Data Lab.

“One of the things we’ve seen among the most successful companies we work with is that they closely align these teams with business goals,” he says. “The way you direct their work and set priorities, the closer you are to achieving the main goal of the business, the more likely they are to [réussir]. As you move more into computing, you are moving away from the primary goals. “

Managers should also think about how they organize their teams when hiring, says Nick Elbrin. According to him, they should ask themselves “what are the basic skills, what will cooperation with others bring”.

“Companies succeed in [constituer des équipes de data science] With people who know basic stats and programming and being trained by people who know devOps or other engineering capabilities,” he adds.

Get management on board

In the meantime, it’s important to ask when professional data scientists are actually needed, compared to tools that claim to “democratize” data science and machine learning.

“It depends on the nature of the problem that you address to the data science and machine learning professionals,” says Nick Elprin. For more classic problems, some automated ML solutions can be effective. If you’re talking about a problem that is unique to your business or at the heart of your differentiation, you need the more flexibility that comes with developing your own proprietary models and using the power of code to express those ideas. “

Finally, developing impactful data science projects requires support from leaders, notes John Thompson. “The real challenge is the change management process at the macro level, it’s not really the data science process,” he explains. To realize the full value of a data science initiative, it is important to make executives understand that “this will ultimately lead to change. You have to be prepared to lead change… If you don’t want to do that, you should probably build a project rather than a program.”

Source : ZDNet.com