Why Data Scientists Wear 3 Hats

Hats come in many shapes and sizes. From the fedora, to the baseball cap to the boater, hats have the unique ability to make a statement. Though not generally known for setting sartorial standards, Data Scientists wear hats too. 

The adage that there is no I in ‘Team’ rings true in the world of data science. Time and again I have heard data science referred to as a buzzword, or the flavour of the day by data scientists, business people, researchers and educators. And it’s not without reason. 

Defining the role however is rather more difficult. As a data scientist working in the real world, not only are you required to understand the underpinnings and applications of statistical and machine learning approaches in the context of your client’s business needs, but you will also need to have a firm grasp of server stack architecture and data base structures. In today’s big data environments, you will also need to adapt quickly to the often unique hardware and software configurations which comprise these environments, as well as be familiar with  fundamental networking principles in order to take advantage of them. 

And there's more...

On top of this, don’t be fooled into thinking that you can make do with simply being proficient in R and/or Python. In truth, adapting to your client’s platforms may require at least a working knowledge of languages like Scala or SAS. I've seen large organisations invest millions of dollars in big data solutions but are unable to make effective use of their data science models until these models are compiled for their particular technology stack.

Wearing many hats

As you can see the data science definition is rather broad. Drew Conway’s classic data science Venn diagram contextualises it as an intersection of domain or industry expertise, maths and stats knowledge, and hacking skills. Individuals who can wear all these hats simultaneously are extremely rare. As a result, the industry remains peppered with diverse skills gaps. 

Of course this isn’t a bad thing. Most data scientists are specialists with some generalist knowledge. We always urge clients to build teams where specialist and generalist abilities co-exist and overlap into well-rounded data science solutions. Such teams will often include statisticians, computer scientists, software developers, visualisation experts, data architects and often business specialists. 

So the next time you're wondering which hat to wear, spare a thought for the data scientist. She strives to wear three different hats every day.