I was reading an interesting article today on the skills required to be a ‘Data Scientist’… All very interesting!
There’s a suggestion that, as you might struggle to get a Data Scientist (note the capital D and S – apparently they earn more than Wayne Rooney), you should aim to get some of these skills in your data team as a group, rather than concentrating them in a single person.
I’d take this one step further. For any number of reasons, you don’t want a data scientist in your organisation.
Any idiot knows you do everything you can to avoid creating over-dependencies in your teams. Similarly, good software architects work hard to remove single points of failure from applications.
Don’t get me wrong… Data Scientists are awesome. Vastly skilled, articulate and intelligent individuals with some punchy geekiness. Without a doubt they’d be a great asset to your organisation… But at what cost?
You need people with great Maths skills, someone who can get data from hugely-complex A to a clean and nicely structured B, someone who knows what to do when your queries take a year to return any data, a bod who can help you understand the technology side of things, someone who knows the business and the specific problem you’re trying to solve… and someone to talk nicely to the non-technical people.
You can have one person doing all this – the fabled ‘Data Scientist’. Realistically, you’re not going to find this sort of person (unless you get one of those people who have decided to try and double their base salary by changing their job titles from ‘Analyst’) so you need to diversify.
Our approach is simple. Get several really bright specialists, and structure a team that allows intellectual freedom and intimate collaboration. If one part fails, you have some backup. Better still, you end up with a truly awesome, well motivated team… Rather than a lone wolf genius.
Think the Borg, but with nicer glasses… and a much better Christmas party.