« How to: Debug in R | Main | Pollution from the BP oil spill »

June 24, 2010

TrackBack

TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a010534b1db25970b0133f1051755970b

Listed below are links to weblogs that reference Why Learn R? It's the language of Statistics:

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

Your french is wrong.
It's "le premier langage" or "la première langue".

Interesting article :)

Saying R lets you formalize things that would be hard to do in mathematical notation just means you need to get better notation. R is a poorly designed language with brittle semantics. Code bases of more than trivial size are less likely to implode than in MATLAB, but that's not saying much.

The linear model example you are so fond of is simple to implement in a variety of languages. The other notations you point out are horrible, and I would never consider them for day to day work. And here we find the only reason anyone uses R: the libraries. Despite the failures of the language, enough people have cobbled enough useful code together in it where it's usually the path of least resistance in statistics to do it in R.

I'm sure the vast library of add-on packages that R has is the thing that attracts the most people to it. Today's count is 3,691 and the number is increasing at close to an exponential rate. But why ARE there so many packages? Because programmers love to work in such an elegant language!

To be fair, the folks at SAS Institute have pointed out that I overstated the case in that quote from my book. While SAS developers do not write SAS procs using the SAS language, they do indeed use their language to write other applications such as Enterprise Miner. Still, I agree with the main premise that much of R is written in R itself and developers turn to other languages such as C and FORTRAN only when forced by computational needs.

Bob Muenchen

See language discussion on Lambda the Ultimate:
http://lambda-the-ultimate.org/node/3726

A more balanced question would be, what are the pros and cons of choosing R?

Off the top of my head:
* R is not written in R.
* R cannot automatically do certain optimizations, because it cannot make certain static analysis gaurantees
* R does all of its operations in memory and does not have good support for enormous data sets
* Although the quality of the graphics for information visualization produced by R are good, they are limited, and extending R to support new visualizations is painful at best (pass me the Advil).

My claim to R fame is being chewed out on the R mailing list by the head of the Vanderbilt stats department.

Is clojure the R we are looking for?

R is a poorly designed language with brittle semantics.

It is obvious programming languages tend to specialize.
C++ for system programming, Java for networking, and VBA for Windows. It is even more likely to expect specialization beyond the IT industry. So R is useful as it currently fills the gap.

"Every language enables thoughts peculiar to the culture."i take your point.learning is a long term.

The comments to this entry are closed.


R for the Enterprise

Got comments or suggestions for the blog editor?
Email David Smith.
Follow revodavid on Twitter Follow David on Twitter: @revodavid

Search Revolutions Blog