With a background in marketing I understood the value of data analytics and business intelligence. However, when I heard about data science for the first time, a new world opened to me. Commercial predictions and decisions made based on data; factual, accurate and aligned to specific business goals. Isn’t that what every modern manager wants? I truly believe that data science will have a major impact on businesses in the coming few years, but who are these people that are able to model the world and change businesses, and what skills do they possess? Sander Harryvan
To apply data science successfully, it is necessary to have a proper and clear business understanding, knowledge of complex mathematical and statistical models, and the ability to develop an infrastructure in which data from various internal and external sources could be collected and connected, and in which algorithms can be ran really quick. Finding one person who possesses all these different and very specialized skills is almost impossible. Not to mention someone who stays up-to-date on all these different domains, since all these landscapes are evolving rapidly. A common metaphor says that when you are looking for such a person, you are looking for a unicorn. These people are hard or even impossible to find, because they probably do not exist.
That is why data science is a team sport. When you want to successfully apply data science in your business you’ll need more than just a ‘data scientist’. You need a team of specialists with complementary skills that strengthen each other, a team that achieves great results together by combining their excellent knowledge on three different domains; business understanding, mathematics and statistics, and data infrastrucures. At Building Blocks, we refer to these team roles as respectively the data translator, the data scientist, and the data engineer. In this article I will give an introduction to these three different roles and I will explain how they strengthen each other.
The most commonly known role in the data science team is the data scientist; according to the Harvard Business Review ‘the sexiest job of the 21st century’. A data scientist often has a mathematical, statistical, or econometric background. He uses his knowledge to develop complex models and algorithms to find patterns in the data, which can be used to predict the future in different scenarios and subsequently calculate the optimal commercial decisions. Isn’t that sexy?
What’s in a name? The data translator is the bridge between the business, the data scientist and the data engineer. Because he has an accurate understanding of business problems and the ability to translate these problems into mathematical models, he delivers the input to the data scientist. This does not necessarily mean that he has the skills to develop complex algorithms himself, but at least he has an accurate understanding of the statistical possibilities and techniques that are invented. He knows what kind of data is available and, more importantly, suitable for specific business problems. This information serves as input for the data engineer, who knows how to collect this data from various internal and external sources. Another important skill of the data translator is that he knows how to visualize the results of the scientist’s work intuitively, in such a way that these are easy to interpret for the business manager in practice.
THE DATA ENGINEER
The data engineer can be seen as the computer specialist of the data science team. With mostly a background in computer science (or something comparable), he knows how to program the solutions into automatized processes. The data engineer is responsible for the data infrastructures on which the scientist’s models run; these have to be quick, scalable, and state-of-the art. The second field of expertise of the data engineer is data wrangling; he knows how to collect data from various data sources and how to connect them. He also cleans and prepares the data for the data scientist, so that he can put his effort and energy in developing the models.
A DATA SCIENCE TEAM
Of course, these different roles are not as black and white as these are presented in this article; in practice there are data scientists, translators and engineers that also have knowledge of other fields of expertise. However, I hope that this article clarified that for one person it is almost impossible to master all these different skills on such a level that results are optimized the best way possible, and that you realize that data science is more than just the sexy data scientist’s job.
For now we have set the scene and know what fields of expertise are important for a successful data science team. In future articles we will dive deeper into these different roles and put more emphasis on their unique skills.