2015 Data Science Salary Survey

Tools versus Tools

Let’s Take a Break from the Money

The first category of tools that’s worth mentioning is operating systems. Windows is still the most widely used (72%), and Linux (50%) is slightly more popular than Mac OS X (43%). Compared to last year, Mac OS X and Windows have both gained 6-7%. Almost everyone uses either Mac OS X or Windows (94%, up from 87% of last year’s sample), and there is a significant overlap between each of these operating systems: all three are used by 12% of the sample (compared to 9% last year), and only 46% use just one of the three.

Specific Tool Usage Rates

Beyond operating systems, we will refrain from imposing our own system of classification.^[6] Tool usage rates on the whole changed little from last year’s salary survey results:

• 68% of the sample use SQL

• 59% use Excel

• 51% use Python

All of the above rates are within 1% of last year’s values.

• R, however, fell from 57% to 52%, although this is only marginally significant (p = .13).

• The new, powerful, and suddenly popular Spark, as well as Scala, the language in which Spark is written, saw large increases to 17% and 10%, respectively.

• Tableau’s share also grew from 25% to 31%.

Aside from R, other tools that are not used as widely by this year’s survey respondents as last year’s include:

• Perl (12% to 8%)

• Matlab (12% to 6%)

• C# (12% to 6%)

• Mahout (10% to 3%)

• Apache Hadoop (19% to 13%)

• Java (32% to 23%)

All of these differences are statistically significant at the 0.10 level. ...