Being a great data scientist is about more than having a head for numbers. It’s about spotting relationships between data, seeing how they solve business challenges, and then conveying this in a way that enables a business to understand what can be done with the information.

For example, if a business in e-commerce wants to know about the background to thousands of individual purchasing decisions, it may also want to analyse the behaviour of the greater number of potential customers who decided not to buy. What did those customers do instead?

Faced with such a challenge, a great data scientist may well be able to find a solution by using analytics techniques that worked for them in another business sector. In this way they must be constantly open to new techniques and new ways of solving problems.

> See also: UK’s data scientists face burnout due to work-related stress

Data scientists also need excellent communication skills. They are not isolated boffins, but increasingly the captain of the team, performing a role that requires them to lead, inspire and communicate at all levels with confidence and simplicity.

But how do data scientists attain these valuable attributes? An examination of their different backgrounds will show that most were already working with data analytics in their previous roles, whether that was as data analysts in the retail sector, for example, or in the academic world, such as in the life or physical sciences.

Indeed most have a computer science element in their academic background or training and a facility for mathematics and statistics. Many, but not all, of those working at the highest levels have doctorates. Alongside this they are able to write and develop code and understand complex statistical factors.

Being at ease with technology must be second nature to the data scientist so that if presented with a new data language or tool, she is not overawed and is perfectly capable of picking it up quickly.

They must also be familiar with the techniques of data ‘munging’ (converting data from one raw form into another in order to make it easier to understand) along with the ability to organise and present their findings with sophisticated visualisation tools.

A science or an art?

In truth, however, all these qualities and skills are just the basics. They will not necessarily make for a great data scientist.

To really succeed, what data scientists need is the desire constantly to question what they are doing and to find out what the next question should be. The great data scientist will want to look for results that were not even considered at the outset.

> See also: Big data: not a magic pill, but an antidote 

And while they may be rare talents, in today’s business world, data scientists cannot work in splendid isolation. To provide insight and return on investment, a high degree of cooperation and team work is necessary. The data scientist is often working with the staff in the analytics or statistics departments of an organisation, so they must be able to develop a common language.

The stress on teamwork springs from the realisation that the uniquely gifted data scientist who can do everything without anyone else’s input, does not really exist. Indeed, the data scientist must frequently act as a bridge between the staff fully engaged in hard-core analytics and the rest of the business.

This is where communications skills become important. Firstly, data scientists frequently have to make presentations early on in a project, outlining the course they aim to follow so they can achieve the full backing they need. Then when the work is done, the data scientist must explain the results of their days or weeks of innovative analytics to senior executives. If they cannot communicate with the business audience and relate their work back to providing solutions to business problems, then its value will disappear, wasting everybody’s time.

Horizontal thinking

In this context, specialisation in a particular industry seems a beguiling prospect – the emergence of the so-called ‘vertical data scientist’ possessed of great insight in one sector. However, in reality this Cyclops approach to data science cuts down on that catalytic ability to transfer successful approaches between sectors. What worked in the telecoms sector or in banking, for instance, might be of relevance to retail. In many cases the analytical techniques are the same but it is the domain that changes.

> See also: Big data should be faster, easier more useful: Jim Goodnight, CEO, SAS

The great data scientist will have this wide experience and will have learned from listening to executives in many sectors.

Above all, the great data scientist has an unremitting focus on providing solutions for the most important problems faced by an organisation. Without maintaining this clear focus, no value at all will be realised from all that hard work, ability and data.

Sourced from Duncan Ross, director of data science, Teradata UK