SAS version for Linux ?

[Marc Schwartz]

Reinhard Sy <hacker-24 at versanet.de> writes:

> Am Dienstag, den 17.02.2009, 16:53 +0530 schrieb Jay Mistry:
>> Is there a SAS (Statistical Analysis System) version available for Linux ?
>> 
> I think NO but you can do all the things with R which is Open Source.
>
> Reinhard

Actually, there is, but it is incredibly expensive if you cannot use
their academic licensing. By definition, Linux versions of SAS are
considered server versions and are priced per CPU/Core. There is not a
single user desktop version of SAS for Linux, as there is for Windows.

SAS is licensed annually, with the subsequent renewals being a
percentage of the initial purchase price. The last time I got commercial
pricing for SAS (where BASE, STAT and GRAPH are now bundled together as
the initial offering) it was >$50,000 U.S. for a 4 Core version for the
first year. A single Core version for Linux was >$20,000 U.S. for the
first year.

Other add-ons (eg. SAS/ACCESS for Oracle, etc.) were several thousand
dollars each.

You don't get to be the world's largest privately held software
company by giving your stuff away...  :-)

I support Reinhard's recommendation of R (http://www.r-project.org/),
which is what I have used for a number of years.

It is available from the above web site via the CRAN network, or
directly from the Fedora repos via yum (eg. '# yum install R').

If you are a SAS user and wish to review a book that can aid with the
transition to R, see:

  http://rforsasandspssusers.com/

BTW, SAS has not stood for "Statistical Analysis System" for quite a
number of years. Their primary business now is in enterprise level data
management and data mining.

If you want to pursue getting more information on SAS, if you really
need it, you need to contact SAS. They have offices worldwide.

A cheaper alternative to SAS, albeit focusing on SAS/BASE level
functionality and not on statistical analysis, is the WPS
Workbench. More information at:

  http://www.teamwpc.co.uk/products/wps/features/workbench

If your primary need is to create/manage proprietary SAS datasets,
rather than needing SAS for statistical analyses, this would be a
reasonable option. 

HTH,

Marc Schwartz