What it’s really like to be a data scientist

Data scientists are in high demand, and while the education they need is grounded in math and computing, soft skills that enable them to support non-technical colleagues and support business outcomes are also essential.

Nearly two decades ago, the television series CSI made crime scene investigator look like a sexier vocation than it was. And while the job title “data scientist” has its own cool factor, it’s important to understand what the work really entails if you’re looking to make it your career.

Data scientists are not a new phenomenon—they just had a different name, says Shaohua Zhang, co-founder of Toronto-based WeCloudData. He’s been doing data science since before it was cool. “It wasn’t even called data scientist when I got started 10 years ago.” Rather, common job titles included “data miner or “predictive modeler.” Today, he says, data science is becoming synonymous with artificial intelligence and machine learning.

The tools are changing, too. Propriety software from vendors such as SAS used to dominate big companies in insurance, banking and telecommunications, says Zhang. It’s still considered one of the best analytics tools available, “but it’s considered a very expensive and traditional tool. A lot of people won’t even have access to that platform.” As more businesses become increasingly data driven, regardless of size, they are turning to open source tools such as Python and Apache Spark, which are what are being taught in universities. “The banks are even starting to move away from those expensive, traditional tools, like Oracle and SAS, and moving toward cloud, open source, and Python.”

Being a good data scientist is more than just about knowing your way around software, however, and it’s not staring at data all day. Communication skills and some business acumen are required, too. “The best data scientists don’t write the most code,” says Max Humber, lead instructor at Bitmaker General Assembly. “The best data scientists solve the hardest and most pressing problems with simple tools.”

Data science is a people business

Unless you’re doing pure research, you’re probably supporting business decisions, so you’ll need be able to articulate your efforts to non-technical people and demonstrate how they contribute to the bottom line, says Fernando Nogueira, a Toronto-based data scientist and Brainstation instructor. “A good data scientist will lie at the intersection of software development, data analytics and business intelligence. How skilled you are at each of the three will determine what kind of data scientist you are.”

Communication skills are indispensable, he says, and the ability to translate difficult technical concepts into ideas other people can grasp and, conversely, being able to translate business requirements to math speak, are also fundamental. “These skills combined are perhaps the most important attribute a data scientist can have.” That’s because your day will be spent interacting directly with product managers, marketers, designers, executives, and clients—not just other data scientists—especially in a smaller organization, says Nogueira. “As you move up in your career the pendulum tends to swing in the direction of more communication and less computation.”

Zhang has found that although many PhDs are skilled at building machine learning models, they don’t know how to explain things to business people. “You need to be able to translate or interpret the complex machine learning models or data sets into very easy language that the business side can understand,” he says. “You need to have the story telling capability. You need to understand all the business use cases.”

There are also certain personality traits that lend themselves to being as successful data scientist. Perseverance is one, says Humber. “Nearly 80 to 90 per cent of your work as a data scientist won’t work. But that’s okay. Because the stuff that works really works.” You should also be a creator, too. While it’s important to keep up with the latest and greatest trends and tools, he says, you often learn more and have more fun when you’re creating.

Nogueira says a good data scientist is curious, pragmatic, and a self-starter. “Without curiosity you won’t spend countless hours digging through data trying to make sense of a weird pattern you noticed.” At the same time, being pragmatic makes for an efficient data scientist, he says. “All models could be made a little bit more performant. All analyses could have the data sliced in yet another way. But knowing when to stop is an invaluable skill.”

It still boils down to the numbers

Although communications skills and business knowledge are critical to delivering value to an organization as a data scientist, a good grasp of mathematical concepts behind statistics, probability, optimization, and machine learning is absolutely fundamental, says Nogueira. “While you don’t need a Ph.D. in statistics to become an accomplished data scientist, simply reading blog posts on the latest fad technique is not going to cut it.”

Analytic skills grounded in mathematics, statistics and predictive modelling are obviously a cornerstone of being a data scientist, says Zhang. “You do need to have some math background knowing some basic statistics and linear algebra.” That being said, you don’t need to understand how to prove things like a mathematician. It’s more important to understand how you can use the tools and theories to apply them to practical data challenges.

Be ready to differentiate yourself

The good news for budding data scientists is that they are in high demand. According to research released by LinkedIn data scientist is one of the most promising jobs for 2019 with a US$130,000 base salary. However, it doesn’t mean there’s a talent shortage.

“It is a busy field with a lot of people with PhD degrees looking for a home,” says Nogueira. “It is easy to get lost in the crowd if all you have to show employers are a handful of online courses.” Like many vocations, having a solid portfolio to show potential employees will go a long way to differentiating yourself, he says.

Zhang equates it to being an artist who has work they can show, such as a hands-on project that shows they can get the work done. “People are competing against each other to become a data scientist, and the bar is usually very high. You have to make sure that you can stand out.”

Interested in learning in-demand data science skills?

Check out Canada’s top-rated learning opportunities for aspiring data scientists.  

Leave a Reply

Your email address will not be published. Required fields are marked *