We’ve all heard it by now: Data scientists have the century’s sexiest job and they’re here to save your business with their big data expertise. Everyone wants to hire one. But what are the chances you’ll stumble across an actual data scientist unicorn who just happens to be a perfect fit for your business? Well, practically nonexistent.
The thing about data science is that it draws from dozens of fields, including machine learning, data mining, analytics and artificial intelligence. Most data scientists hold graduate degrees in computer science, but many of them come from backgrounds ranging from electrical engineering to biology. In other words: No one data scientist possesses the breadth of skills needed for every data science position. Believe it or not: there are no unicorns.
And, of course, competition for data scientists is fierce. You may know that McKinsey & Company has predicted that the U.S. faces a shortage of between 140,000 and 190,000 analytics experts by 2018. Meanwhile, the number of open jobs is on the upswing -- interest in data analytics talent increased by 15 percent from 2013 to 2014 alone, according to Indeed.
So how can you hope to make a hire that fulfills your requirements from such a small pool of qualified talent? It starts with actually knowing what those requirements are.
It seems simplistic, but data scientists rarely have equally strong backgrounds in areas like statistics, time-series analysis and database know-how. The secret is finding a candidate with the right mix of skills that match your business goals. For instance, if you run a retail business, a time-series analysis is a great way to understand patterns in product sales over time. On the technology side, you want candidates who have experience using the tools that are central to data science, such as Python, Spark, or R.
As with any job, it’s pretty easy for a candidate to list the right bullet points on their resume. Luckily, in data science, it’s also pretty easy to find out if a candidate isn’t as savvy in certain techniques as he or she claims to be. An initial phone interview with a hiring manager who has a technical background should serve as a quick check. Beyond that, there are several ways to test the depth of a candidate’s knowledge.
First, a real data scientist is most likely going to have a collection of data-related passion projects or academic work that will tell you way more about his or her skills than any take-home problem you might assign during the interview process. Ask for a link to a project on the candidate’s GitHub profile or a copy of a published paper. But don’t stop there -- during the interview, ask hypothetical questions based on scenarios a data scientist might actually encounter at your business. How your candidate approaches the problems will tell you a lot about his or her critical thinking process and what tools he or she is most comfortable with. You might even want to hand over some sample data and ask for a quick analysis.
But now, a word of warning: don’t get completely bogged down in testing for technical skill. Data scientists have unique roles because they straddle the line between analytics and business strategy. If they lack communication skills or curiosity, they won’t make the effort required to uncover business-changing insights and communicate them in a clear, concise way to your decision makers. And if that’s the case, you’re paying out a six-figure salary -- averaging around $120,000 for non-managers -- for little to no return on your investment.
There are some ways you can ensure this kind of scenario doesn’t play out. Ask candidates to explain technical concepts such as the difference between logistic and linear regression, to a hiring manager with a non-technical background to see if they take their audience into account. Employ behavioral interviewing techniques, where you ask questions about a candidate’s past performance in particular situations. Involve multiple parties across all departments to ensure that your candidate is a good fit for your culture.
What sets hiring a data scientist apart from any other employee is that so much rests on a data scientist’s shoulders: The promise of profit has 75 percent of companies planning to invest in big data by 2017, and the range of industries now employing data scientists is staggering. Unless you want to be left behind, now is the time to make a data scientist hire. Just make sure it’s the right one.