April 30, 2018 by Luke Mason
Last year, we took an in-depth look at the demand for data scientists—those tasked with collecting and crunching huge amounts of data and turning it into useful information. We found that data science is an interesting combination of math/stats skills and computer science/software skills inside a particular subject-matter domain or industry (such as retail, marketing, finance, or national defense/intelligence). We also discovered that the knowledge, skills, and tools in data science vary from region to region, based on the local industry mix and major employers.
The more we explore our data on this fascinating occupation, the more insight we uncover. Here are just a few new observations we took from our job posting data. We use some cool Sankey diagrams to visualize.
First, when we analyze all job postings that specifically use the job title “data scientist,” we see that postings have grown over 600% from 2013 to 2017. Demand is extraordinarily high, and it isn’t going away anytime soon.
Note: Companies sometimes post for job titles besides “data scientist,” which means the growth in demand could be even higher than 600%. We continue to see a proliferation of industry-specific titles: information research, financial analyst, business analyst, market researcher, etc. But for our purposes, we’re restricting our analysis to the data scientist job title.
Data science jobs are in-demand everywhere, but three MSAs lead the way with the greatest number of job postings: New York, Washington D.C., and Seattle. In these places, a handful of industries drive the demand for data science skills. In New York, the skills relate to finance, whereas in D.C. they concern national defense and intelligence, and in Seattle, Amazon’s rapid growth is driving fierce demand for a broad array of data science skills.
Back in 2013, Microsoft created the most postings for data scientists. In 2014 and 2015, other companies like IBM and Booz Allen joined the scene. Then in 2016, Oracle and Amazon started to dominate, with Amazon becoming the clear leader in 2017. Anthem is also prominent, while IBM has fallen far down the list.
(Keep in mind that more postings doesn’t mean more jobs. Companies might publish a greater number of postings than they have open positions. The number of postings simply indicates greater ferocity in marketing those open positions.)
Now let’s take a broad look at how the skill needs have changed in data scientist job postings across the U.S. In the chart below, you can see the evolution of skills from 2013 to 2107.
Back in 2013, the top skills mentioned by employers were data mining, research, statistics, and Apache Hadoop. Today, statistics is still in high-demand, but machine learning tops the field. Much of the demand for machine learning is due to Amazon’s mega influence on the marketplace.
It’s also interesting to note the combination of this older, tried-and-true skill (statistics) with the newer, emerging skill (machine learning). Emerging sectors like data science are often characterized by such pairings of old and new skills (as we also observed with modern manufacturing.)
This is because new technologies and techniques are often quickly adopted by companies looking to innovate on top of proven methodologies. When statisticians gain next-level processing power and tools that allow them to crunch far larger datasets, you get entirely new areas of work like machine learning.