By using website you agree to our use of cookies as described in our cookie policy. Learn More

TDWI Upside - Where Data Means Business

Is Your Data Science Team Sufficiently Diverse?

Data science is critically dependent on contributions from a broad spectrum of people -- from business analysts and data engineers to business domain experts and data scientists themselves.

Data science is inherently interdisciplinary, which means it's critically dependent on contributions from a broad spectrum of people -- from business analysts and data engineers to data architects, business domain experts, and data scientists themselves. The upshot is there's an important question you should be asking yourself: is my data science team sufficiently diversified?

The Value of Diverse Backgrounds and Skills

In a wide-ranging presentation at Data Science ATL, a 2014 conference held in Atlanta, data scientist and all-around polymath Paco Nathan touched briefly on how data science is inescapably interdisciplinary. As Nathan sees it, data science is as much a problem of managing -- and innovating in the context of -- people and social issues as of using the tools of statistics, math, and machine analytics to wrest insights from data.

"[Data science is] inherently interdisciplinary. It's inherently about the people and the social issues in an organization. That's a lot of what you have to knock down [i.e., overcome], along with the math," Nathan said. "Data science is ... a lot about teamwork; it's a lot about bringing in people who have a diversity of different backgrounds and skills."

The composition of a data science team with a Web start-up in Silicon Valley won't look much like that of a manufacturer of consumer packaged goods based in the Midwest.

However, both teams will have something in common: a diversity of perspectives and skills.

Angela Bassa, senior manager in data science with a prominent provider of energy intelligence software, sees this diversity as a critical aspect of the work her team does. "I'm a big believer that ... a team has to have people who have one another's blind spots covered. A well-functioning team is ... going to have people who have a lot of expertise in all of the relevant domains [for the business]," she explains.

Domain expertise is potentially a huge category, however. In Bassa's case, it includes people with mechanical and electrical engineering skills -- along with physicists. "If we didn't have people who truly understand how a facility operates and how billing operates, we wouldn't be able to tackle these problems because we wouldn't be able to articulate what the problem is," she says.

Translating from Math-Speak

Everyone expects a data scientist to have strong math skills. However, Bassa says language skills -- more precisely, the ability to make extremely complex concepts lucid and understandable -- are no less important. Think of this person as an epistemological translator of sorts.

He or she must be able to work with -- and translate to and from the language of -- engineers, architects, developers, statisticians, analysts, data scientists, business stakeholders, and others.

"We need people who are well-versed in the language of math, yes, but we also need people who are good with translating from that [complex math] into ordinary language. Math lacks the ambiguity that language brings, but language is also very important [because] we have to be able to communicate with our customers," says Bassa, who has a mathematics degree from MIT.

"My team is the data science team; we come up with the nerdy math. [However,] if I can't put that into production, my customers can't take advantage of it. I need to be able to work with developers and designers and testers and people who have very different [roles and] responsibilities," she says.

Cross-Pollination: Diversity by Another Name

Diversity of perspective isn't just a function of one's background or domain expertise. Bassa uses her own experience as an example. "Working in finance [out of college] taught me lots of things, working in hospitality [as a scuba-diving instructor] taught me lots of things, coming back [to business and] consulting taught me lots of things.

"[Experiences with] all of these different industries allow me to cross-pollinate and pick different approaches [from] different industries," she says. "That's one great thing I bring to the table. On our team, we have people with decades of experience exclusively in energy. It helps that I can bring in these crazy ideas from other industries."

Nathan touched on this same issue -- cross-pollination -- in his presentation at Data Science ATL, buttressing his claim that data science is "inherently interdisciplinary" with a quick history lesson. He explained how innovation in cybernetics, biology, and other fields helped inform (and in several cases, critically influenced) the development of machine learning technologies (such as neural networks) and AI.

Cross-pollination of this kind presupposes a wide-ranging intellectual curiosity. Certain techniques became portable and applicable to other fields because curious, imaginative people learned about them and saw the connections.

Intellectual Curiosity Both a Feature and a Bug

As Bassa sees it, successful data scientists and other like-minded people -- "nerds," as Bassa, a self-described nerd, puts it -- are by nature intellectually curious. This is on balance a great thing, she says. It does have a few drawbacks, however.

Bassa invokes an example from the Pixar movie Up: Doug the dog and his tendency to be easily distracted. When Doug sees a squirrel, he forgets whatever he had been doing. She and other nerds operate in much the same way, Bassa says.

"There's so much interesting and fun development happening that it's easy to forget what our specific goal is. Of course it would be fun to play with neural networks, but most of the low-hanging-fruit problems don't require a neural network to solve. It's very easy to become enamored with a methodology and to find out ... no, it's [too] overengineered for our little problem that we have scoped out very clearly," she concludes.

"I think intellectual curiosity is amazing, and table stakes to be a good data explorer; I just also know that even curiosity has a potential negative aspect that needs to be well-managed," she says.

TDWI Membership

Accelerate Your Projects,
and Your Career

TDWI Members have access to exclusive research reports, publications, communities and training.

Individual, Student, and Team memberships available.