Welcome to Analyzing Analytics, a new blog that examines the burgeoning world of Big Data analytics and explores how companies across a multitude of industries can leverage data to improve business performance. In today’s first post, we look at the insatiable demand for data scientists and identify the key attributes of a solid and effective scientist.
This past year, the thirst for big data only intensified with greater recognition of the positive impacts it can have on a wide swath of business and society. There is little to suggest any abating of this as we move into 2015.
One shift that I do expect in 2015 and beyond is an expansion of focus from just the technology layer to the analytics layer as well. As I’ve seen on too many occasions, those who rely solely on data collection technologies, systems, and platforms tend to fall short of their expectations. It’s the analytics piece that is crucial, and real value creation greatly depends upon the capabilities of the individuals – data scientists – who have the ability to parse vast amounts of data, apply complex mathematical techniques and arrive at actionable, usable and purposeful outcomes.
As you might expect, people with this skill set are in high demand. In the UK, for example, approximately 56,000 big data jobs will be created in the each year until 2020, pushing the rate of job growth in data and analytics to 160%. Universities around the world are scrambling to train a veritable army of data scientists to address the growing need. Northeastern University, for example, recently announced new programs designed to train data scientists. Northwestern offers a Master of Science in Predictive Analytics (MSPA) program dedicated to data science training. Prospective scientists, as one might expect, are looking to capitalize on what they see as a white hot job market, as evidenced in part by the fact that enrollment has nearly doubled in a 2-year-old analytics master’s program at the GW School of Business.
The demand for people with these skills and the existing shortage means that many companies are likely to make decisions to hire employees or vendors quickly, leading to the possibility of errors in judgment and mistakes that could set the hiring company back. All of this begs the question – what makes a good data scientist? I believe quite fervently that the one who only has skills in “data” and “science” does not necessary become a good data scientist. More specifically, being good at “data” and “science” are necessary but not sufficient conditions to being a good data scientist.
So to help companies in desperate need of data scientists, I’ve identified the top traits that hiring managers and executives selecting appropriate analytics vendors should seek:
Understanding the Business Context before jumping into the Science
A great data scientist will invest the time required to understand the context of the business. Most times, there isn’t a lot of clarity as to the exact problem. A good data scientist will “co-discover” the problem with business partners. They will ask clarifying questions, discuss other related problems/opportunities and share multiple approaches. They will then arrive at a broad consensus on the specific opportunity to pursue and the high-level expectations from the initiatives. All this gets done before the data crunching and statistical modeling work even starts.
Comfort with Imperfectness
Math is specific. Database code is precise. The real world is messy. Data is scattered. Not all data is accurate. Some data issues can be fixed easily, other issues can be fixed with a good deal of effort, yet other data issues are next to impossible to address. Great data scientists are aware of the messiness of real world data, and they take it in their stride. They are adept at pushing things forward even when perfect data doesn’t exist, because perfect data is only present in textbooks.
Drive towards results
Businesses drive towards results. While people might be comfortable with “cool” insights and thoughts in the short term, ultimately they will respect a data scientist who creates tangible value. A great data scientist is aware of this. From the beginning, they have their eye on how the final recommendations will get implemented, and work backwards from there. They understand that the final users of their work – whether humans or automated systems – come with their own unique constraints & nuances. Hence, they plan their approach in a manner that ensures that recommendations gets implemented and real value gets created every single time.
Effective Communication, especially with business managers
Big data is not a technology or statistical initiative – it’s a business initiative. A great scientist understands this key difference. They know their organization is adopting big data with the ultimate goal of improving business metrics and, therefore, they communicate in a similar fashion. This means less “data” and more “insights and recommendations.” This means making it easier for a non-data scientist to understand, assimilate and react to findings. This means talking less about the “how” of the analysis and focusing more on the “so what” of it.
Hunger to learn
The world of analytics and big data is changing rapidly. New technologies, new use cases and new platforms are springing every single minute. The skills that someone has right now will only go so far. A great data scientist keeps pushing their own envelope and the envelope of her organization. They try out new data management technologies, evaluate new use cases and familiarize themselves with less used statistical algorithms. A great data scientist understands that sustained success requires a continuous drive towards learning.
So there you have it – key qualities to look out for in a great data scientist that are not centered around either the “data” or the “science” elements. What has your experience been?