Published: Wednesday March 22, 2017
Continuing to share the stories of GBG’s talented people, we asked our Senior Data Scientist, Dr Ian Hopkinson to give an insight into the world of data science…
Can you explain data science to the uninitiated?
A colleague asked me “How do I become a data scientist?”
My reply was: “First become a physicist, and then cross out the “physicist” on your name badge and write in “data scientist”!”
I got the impression they didn’t consider this to be an entirely useful answer.
For me being a data scientist is an extension of my natural enthusiasm for data. Physics is about collecting data from the physical world and drawing conclusions or building models on the basis of that data. Physics is about prediction: How fast will the apple fall? Data science is about generalising that world view to other types of data; it might be the physical attributes of houses in the UK, or the behaviour of users on a website or in a store, the movements of share prices, or the catalogue of the Science Museum. The basics are the same: collect, visualise, model, predict.
What are the key skills necessary?
In practical terms a large part of my job is finding the data I need and rendering it into a state in which I can work with it. To the traditional skills of the physicist I’ve added those of a software developer – in particular source control, testing and more recently deploying systems into the Cloud*. Explaining clearly what I’ve done to the audience in front of me is also important, this is a skill I picked up whilst teaching students and also as an industrial research scientist, explaining my work to a project team.
*My former colleagues used to deploy actual measurement devices into real clouds with an aeroplane!
What do you believe makes a good data scientist?
The reason I’m a good data scientist is that I’m very concerned with what the customer wants, and hence the revenue generating value of the work I do. Sometimes this means being a bit brutal about the techniques I employ. Every data scientist will love to build a machine learning system and so forth but if actually you get almost all the value from a simple “rule of thumb” then really that’s what you should do.
What’s been your proudest achievement at GBG?
First paying customer for the Location Intelligence dataset!
Ian has been a member of the GBG team since November 2015. He’s helped develop many of the attributes in GBG’s ground-breaking Location Intelligence solution. Learn more about it here