On 27 August 2007 at 13:19, Andy Bunn wrote:
| Is there a reliable (for some definition of reliable) estimate of how
| many people use R or have downloaded it? Say an order of magnitude
| estimate? I would like to mention this in the introduction to a paper
| I'm writing where I encourage R's use.

With a salt mine rather than a mere grain of salt, you could consider the
'popularity contest' results from Debian and Ubuntu.  Briefly, it's an
'opt-in' service that submits the list of installed packages, anonymoysly, to
an aggregation address.   

Using Ubuntu's large install base, we get from http://popcon.ubuntu.com that
eg the GNU bc package in their 'main/math' section is installed (and
reported) 174452 times (based on http://popcon.ubuntu.com/main/math/by_inst).
As this is a mandatory package, we can use this as the baseline.  On the
other hand r-base-core (from the optional universe/math section) is installed
2638 times (see http://popcon.ubuntu.com/universe/math/by_inst).
So you get 2638 / 174452 or around 1.5%.  You now need to figure the various
self-selection biases and correct for those...

As a check, for Debian (with popcon.debian.org as the base URL), we get 1686
/ 54221 or around 3.1% (using file http://popcon.debian.org/main/math/by_inst).  
So even this crude measure has a large amount of variability.  So I still
invoke fortune(43).  

But if you insist, you could use Microsoft's recent 'one billion PCs'
estimate (news.bbc.co.uk/1/hi/sci/tech/2077986.stm), apply the percentages we
laboriously derive above and call it 15 to 31 million....

Dirk

-- Three out of two people have difficulties with fractions. 
______________________________________________ 
R-devel@r-project.org mailing list 
https://stat.ethz.ch/mailman/listinfo/r-devel