The most popular programming languages on StackOverflow

by Andrie de Vries

Last week, IEEE Spectrum said R rised to #6 in Top Programming languages. They use a weighted methodology of 12 factors to compute their score. Among these factors is the activity on social programming websites, including StackOverflow and Github.

I recently used data.stackexchange.com to query the total number of questions on StackOverflow using the R tag.

It is easy to extend the query to include all of the top 10 languages (according to IEEE Spectrum) and see if the StackOverflow activity tells us anything interesting.

This plot shows the monthly total number of questions in each tag on StackOverflow, since 2008. JavaScript and Java tops the list, with R in position #8, roughly on par with C.

 

There is a fair amount of noise in the form of monthly fluctuations. The underlying trends are slightly easier to observe by fitting a smoother through the data:

From this it is apparent that R is increasing its rank compared to the other languages. During the past year, it has been catching up to C and should overtake soon to be in position #7. By comparison, during most of 2013, R was in position #9 only.

<edit>

I've had several requests, from Dirk Eddelbuettel and Hadley Wickham, to include a plot with a log y scale.  Here it is.  It seems to me that the R-tag has a steeper slope than the other tags, indicating faster growth.

 

Also, Gabe Becker asked if this includes question on CrossValidated. The answer is no, this is only questions on StackOverflow. On CrossValidated there are an additional ~400 R questions every month. In other words, about 10% of the volume on StackOverflow.  This will make a small, but not hugely material difference.  Note that this also doesn't include tags on any of the other StackExchange sites.

</edit>

 

Analysis

The IEEE spectrum study uses many other factors, including search rankings and job postings.  For R to jump to #6 in their study means that these other factors also play a role.

Nevertheless, it is interesting to compare the numbers on a single, very easily accessible metric.

Caveats

I didn't try to include any of the other languages in the StackExchange data query, using only the list of top 10. Overall ranking might change in my analysis with a more comprehensive list of tags to search for. (You can change the underlying query and add more tag names)

The code

Here is the code to make these plots:

 

Comments

[Hadley Wickham]

Can has log y scale?