3 Critical Competencies to Look for When Hiring a Data Scientist
Enterprises used to focus on their data in hindsight. But with the rise in complexity of analytics, increasingly difficult problems require greater talent to solve them. Questions are no longer confined to retrospective examination. Instead, data-savvy enterprises ask “what is going to happen” (predictive analytics) and “what should happen and what do need to do to get to that end?” (prescriptive analytics). These questions are hard and require a special breed to take them on — hence the rise of the data scientist. Even among the tens of thousands of trained data scientist, hiring a good one is a challenge. Any data scientist you feel may be a fit for your organization must have at least three core competencies.
- Data Management. Exabytes and petabytes are now the common measure of data repositories. Data scientists must thus understand and be able to manage data. Most advanced analytics involves finding relationships across data sets, so data integration is a must. Experience and ease working with unstructured data is also critical.
- Analytics Modeling. Analytics modeling is the process of formulating ways to extract insights from data. The use of business intelligence and predictive analytics is now commonplace. Advanced algorithms are applied against massive data sets to infer actionable intel in areas as diverse as supply chain management to aviation. This diversity requires the ability to recognize the appropriate analytics techniques for the data set and challenge. Domain and industry knowledge are also helpful.
- Business Analytics. This is the most challenging domain for data scientists who are not trained in business. They are asked daily: “What business goals and questions need to be answered? What business decisions will their analyses affect? What constraints are there?” These are critical questions that determine how an enterprise must infer actionable intelligence from its data. Can a data scientist be trained to be a company- or industry-specific business analyst? Sure, but not overnight. Another critical question is whether the CIO will share his C-suite knowledge downstream so as to facilitate answering these questions. On that front, McKinsey and the National Academy of Sciences assert that using analytical tools should be easy and even fun.
- McKinsey writes that “sophisticated analytics solutions . . . must be embedded in frontline tools so simple that business managers and frontline employees will be eager to use them daily.”
- The National Academy of Sciences has referred non-judgmentally to these employees as “naïve users” for the purpose of impressing upon corporations the need for Big Data to be a business line imperative supported by IT, not vice versa
When The Going Gets Rough, Look Out Your Window
It may be unrealistic to think that a data scientist candidate–or even a team of them–will possess each of the these three competencies. Likewise, a data scientist new to a company is likely to realize pretty quickly that he doesn’t have the human capital to solve all the problems that he has identified with the help of analytics. There’s too much data, not enough time, and almost always a shortfall in analytics horsepower. This makes it nearly impossible to cull the seminal information from data sets. In those situations, data scientists and their supervisors (e.g. CIO, CTO) should look outside their own four walls and follow the admonition of Bill Joy. An American scientist and co-founder of Sun Microsystems, Joy famously said that no matter where you work, the smartest person in your field works for someone else (Joy’s Law). This is not a criticism of any data scientist. The same applies in any field. Data scientists, however, have extraordinary opportunities to adapt and multiply their own internal talent base through the use of crowdsourcing. Doing so is critical given the time pressures of their internal initiatives.
Hundreds of thousands of data scientists await the call to complement those core competencies and solve problems in ways that will make their clients internal champions, if not outright heroes.
In the meantime, hire patiently and intelligently. Just any data scientist is rarely the data science you need.