The latest insights from your peers on the latest in Enterprise IT, straight to your inbox.
Leigh Martin, Senior Director, Dynamic Science Labs, Infor
Big data analytics has become mainstream. Investment in the infrastructure for collecting and managing large volumes of structured and unstructured data tops the list of IT priorities in many companies. As a result, a key hurdle exists and is even growing in intensity—the scarcity of people with the right skills to turn data into insights.
To bridge the current gap between the supply and demand of data scientists, many new advanced academic programs in data science and analytics have been recently established and many computer scientists, statisticians, and operations research professionals have gone through additional training in new data analysis tools and techniques. But these newly-minted data scientists, while conversant in the latest data science languages, also need to be familiar with the language of the business, have adequate understanding of IT issues, and be able to link insight to action.
Here are the 5 skills that define a successful data scientist in any business setting:
Great data scientists think and act like scientists
Scientists are not only highly curious and inquisitive but they also have a feel for what is the right question to ask. Posing the right question is how you start a successful journey of discovery.
After asking the right question comes the task of looking for and finding the right data to answer it. This is particularly important in the big data era where many companies have over-indulged in the collection of data. Scientists understand what type or source of data may answer a specific question and know how to avoid drowning in (or distractingly fall in love with) what’s available. This also includes the ability to aggregate raw data, “bubble it up” so to speak, especially when the data on its own (e.g., RFID signals) is not meaningful enough.
Crucially, scientists see the forest from the trees. They understand what model may work with a specific set of data and how to develop it in order to derive insights that may answer a specific question. Scientists model the world, and perhaps most important, they love to experiment with it—applying and adjusting different types of models, pitting one hypothesis against another—and doggedly try and try again until they “get the science right.” Great data scientists are good at applying the scientific method to business problems.
Great data scientists earn the trust of the business
These days, technical expertise by itself does not cut it and we expect data scientists to have business acumen—knowing how the business works and knowing how to interact with business people.
Even on large projects, where the team includes a representative of the business, typically someone with knowledge of analytics who is responsible for explaining the results to other business executives, having business-savvy data scientists really helps smooth the process and ensure success.
Most important in this regard, is the ability to earn the trust of the business. The goal should always be to make data science consumable and relevant. The Googles and Facebooks of the world got the consumers of the work of data scientists to expect algorithms that integrate, almost seamlessly, into our everyday lives. So making sure that the end-users of our work are convinced that the science is good and credible is a very important part of what data scientists do. Data scientists are engaged in earning the trust of the business throughout the life of the project and beyond.
Great data scientists know how to build bridges between IT and the business
In any data science project, you need IT’s help—they must contribute their expertise in data management, security, and the supporting infrastructure. But if you work separately with IT and with the business, you end up taking twice as long and investing twice the effort.
Data scientists must understand IT issues and what keeps the CIO up at night. With their sensitivity to the challenges of IT, coupled with their knowledge of the business and its requirements (see above), they serve as a crucial link between IT and the business. Great data scientists know how to translate the language of the business to the language of IT and the other way around. They are able to turn the user’s need into a business requirement and then turn it into a technical specification, one that would work in a given IT environment.
Great data scientists can put themselves in your shoes
It is customary to highlight “communications skills” and “storytelling skills” when discussing what is required of data scientists in our team. But being able to explain the results of data analysis to someone without training in data science is only one side of the coin. The other side of a successful data scientist is listening skills or more broadly, empathy.
Successful data scientists know how to engage with business users—discussing their business problem, explaining how they would approach it mathematically, illustrating what works and what’s not working, listening to their questions and suggestions—and all of these require empathy, putting yourself in someone else’s position. It’s an iterative process, a dialog, in which they teach you about their specific business and you teach them about data science.
Great data scientists never stop learning
Data scientists today perform multiple tasks. Years ago, we used to have more structured roles. We had a data analyst mining the data, a scientist developing the algorithms, and the business person defining the business problem. Today, these roles start to meld together.
No matter what specific discipline you were trained in, once you take on the role of a data scientist you do everything—code, develop algorithms, understand the business process, clean and massage the data. Most important, you constantly keep abreast of all the latest developments that are relevant to your work. You go to conferences to learn new tricks, talk to your peers, and meet new people with new perspectives (who you may later recruit to work with you). And you take advantage of all possible resources and sources of new knowledge—online courses for specific tools, for example, to say nothing about books. Yes, books—at our Dynamic Science Labs we maintain a library with books that are relevant to our team’s work.
Constantly learning, experimenting, exploring. Great data scientists always push the envelope of the science of data analysis.