TDWI | Training & Research | Business Intelligence, Analytics, Big Data, Data Warehousing

The Best Job in America: Data Scientist

Glassdoor recently released their review of the 25 best jobs for 2016. At the top of the list is the vaunted data scientist. What does it take to be a successful data scientist?

• By William McKnight
• February 25, 2016

Glassdoor recently released their review of the 25 best jobs for 2016. The ratings were determined by three key factors – number of job openings, salary, and a career opportunities rating. At the top of the list is the vaunted data scientist. Congratulations to all the data scientists and those who aspire to be one.

One of the most interesting aspects of the ranking to me is that (at least a low-level form of) data science is something that we've been distributing to the entire data consumption audience for years. For there to be a dedicated job, and the number one job at that, speaks to the extreme importance of data.

I sat down with data scientist Mark Schwarz, VP of data science at Square Root, to discuss data science, the data scientist, and this study. Mark started with defining the data scientist as someone who turns information into actionable next steps -- a forecast, a cleaner classification, or a statement.

One thing that is evident is that the successful data scientist must be able to communicate. It's not a role for someone in a "back room" doing analysis and coming out of the dark monthly. The data scientist actually needs to be very interactive. A data scientist will spend a considerable amount of time communicating. For example, in the Kaggle competitions, competitors spent 10 percent of their time on average on model construction and 90 percent on human interaction in setting up the problem. Feature creation and feature selection are very interactive.

As for skills, Mark said R and Python are the main skills in demand, but the models did not have to be complicated to be effective. Simple linear models, clustering, and outlier detection and removal would take many companies quite far. Those are some of the technical skills required, but I found it quite insightful when Mark said the main skill required is the ability to structure a problem. Creative skills are essential.

Organizational change skills are necessary as well. Mark cited an example of a retail client needing to balance their natural drive for high-dollar tickets with the objective of customer retention, which the data science determined needed to be more prominent in the company's behavior.

Engaging data science is most effective when it lacks specific desired outcomes and allows for the science, and the business, to go where the data takes it. We find on our data science engagements far less specificity in the desired outcome and a greater openness and desire for change. This is key to setting up the data science project for success.

The high ranking in Glassdoor's report is not a fluke, and it is quite likely that the data scientist will remain top of such lists for years to come. The human need to interpret the results and play the role of the human counterpart to the science will remain strong. Data science will permeate business activities and it is likely the data science role will evolve into many levels in most organizations in addition to permeating roles everywhere in the organization.

There are still many unknowns, and we have just begun to codify the role.