Friday, January 9, 2009

R Rocks

I use R pretty much constantly, so I was pleased to see it getting some attention in the NYT (see also Mailund, Hawks, and Dechronization). There is also a followup post by the article's author here.

One of the best things about R is that it is freely available, this is a huge boon to any kind of statistical research as most other statistical packages are prohibitively expensive putting them out of the reach of many researchers. Because of this the R community is really great. I often find that the problems I come across when coding have already been solved and code is available freely online.

The idea expressed in the article by the SAS rep. that some how propriety software is safer than open source is laughable (and really just makes them look silly). I would far rather work with something that thousands of statisticians spent every day picking apart the code, than a blackbox from some corporation that has a strong interest in saving face.


Steve said...

Yes, R does rock for some people. For others, SAS rocks. It's a big analytic world and there's room for everyone.

As a SAS marketer I suggest that if you're really interested in this topic, you check out a followup post by SAS' Anne Milley who was quoted in the Times. It's at

G said...

Thanks for your comment Steve, and for pointing me towards the SAS blog. I agree that it is a big analytic world, it is a shame that the brief quotes in the NYT article do not particularly carry that view across well (I can well understand how this can happen in a longer interview). I would be interested to know whether there is any actual data about the reliability of commonly used R vs SAS packages, as it seems to me that misleading results are usually noticed by experienced users and the bugs dealt quickly with (as usually occurs in all open source). It seems like the tech. support is the major clear advantage of SAS.

G said...

A company might choose not to release software that has not been accurately tested, and thus a naive user can use the company's threshold as the golden-standard for what's reliable...but for those users who are able to make such judgments themselves...if they can screen out the poorly developed packages...there's a much higher its best, the open source community can test packages more than anything a company with limited budget can do.

Anonymous said...




Anonymous said...

Great one!!

hydie_mojo said...