The popular image of data scientists is very much a bit of a geek or nerd. However, that’s not strictly accurate. It may not be – as Harvard Business Review famously suggested – the sexiest job of the 21st century. But it is a long way from the perceived world of the (usually male) computer-focused nerd.
Data science is all about creating value from data. It allows us to better understand relationships, uncover hidden business drivers, predict future trends and behaviour and make better decisions. We call this intelligent decisioning. However, what data science skills are really needed to uncover all the gems in data? What should companies look for when they recruit data scientists?
Uncovering insights on data science
Recently, I interviewed Professor Andreas Brandenberg, Director of the master's program in applied information and data science at Lucerne University of Applied Science and Arts, and Umberto Michelucci, Head of the AI Competence Centre at Helsana, a leading health care insurer in Switzerland. The interview was a virtual panel talk at SAS Forum DACH. COVID-19 meant that the approach was a bit improvised, with all of us working from home. However, it also felt very timely because, of course, COVID-19 will have huge implications for data science.
We know that today’s expectations for data scientist skills are both broad and deep. Many companies want the full range of skills, from gathering relevant data to generating value by embedding analytical models in business processes. These skills, therefore, include both technical and business knowledge, project and process methodology, data science tools and languages, and knowledge of the IT environment. Andreas Brandenberg has also noticed a high demand for softer skills.
Job skills for data scientists
First, Andreas suggested, data scientists must be able to ask the right questions about both technical issues and business-related themes. Data scientists may have a nerdy image, but they also need considerable creativity and empathy. They must act as a bridge between domain experts. They need to understand and interact with both the IT team and the business side, to make the whole operation perform. And they should be able to explain and “sell” the approach to both the management and the team. Data science should not be a “black box” or some kind of alchemy. So the team needs to understand, buy into the approach, and then work together to deliver.
Umberto confirmed this view and added that data scientists must be team players, able to work easily with all the different domain teams. They must also be storytellers, able to explain the approach and process in words that the audience understands. Umberto also felt that they needed to be problem solvers. This is because the real issues in a production environment are often different and more profound than the theoretical ones in worked examples.
Developing data scientists
It is clear that the expectations of data scientists are vast. The whole environment is extremely interesting. There is lots of potential for data scientists to move in different directions, including marketing, communication and business development. Andreas suggested that the future controllers are likely to be data science-based. Data science will diffuse into many other professions, even unexpected areas like history – where text mining is omnipresent already.
This potential is also the reason why the Master of Science in Applied Information and Data Science degree program at the University of Lucerne is fully booked. This program includes an elective SAS specialisation module. Andreas suggested that there are three main reasons for the degree’s popularity. The first is employability. Many companies use SAS to create insights and generate value. The second reason, which is closely connected, is that SAS Viya is a professional full-stack data science platform, which also integrates open source programming. This is now essential learning. Finally, SAS offers a comprehensive academic program including SCYP, with software, e-learning, workshops and certifications.
The comprehensive profile required of a data scientist is likely to further deepen and widen. We can already foresee that data scientists will require IT skills, soft skills, statistical knowledge, analytics tools, processes, methodology, and interdisciplinary exchange knowledge and skills. - Umberto Michelucci
I asked Umberto one final question: How will be the expectations of data scientists’ skills change over the next two to five years? His view was that it was a good idea to invest in training and certification because expectations were only likely to grow. The comprehensive profile required of a data scientist is likely to further deepen and widen. We can already foresee that data scientists will require IT skills, soft skills, statistical knowledge, analytics tools, processes, methodology, and interdisciplinary exchange knowledge and skills.
An important part of the team
Value creation is a team effort, and data scientists are an important and sought-after link in that team. It is even possible that data scientists may be the fabled unicorns. There is, however, no doubt that they are considerably less nerdy than you might have thought. Perhaps their most important skills are not necessarily technical, but the soft skills – making the profession a wise move for many with a far from conventional "techie" background.
The recorded discussion is available on YouTube (in German, with English subtitles available): Expert Panel: Closing the Skills Gap in AI.
