AndyBolton
Posts: 2
Joined: Sat Dec 24, 2011 3:29 pm
Contact: Website

Re: Statistics, Analytics and Modelling...

Fri Feb 03, 2012 9:47 pm

Is anyone else interested in seeing if we can get WPS from http://www.teamwpc.co.uk to work on the pi?

I don"t think SAS would play ball, but I"ve already sent an e-mail to Team WPS to see if they would be interested, as the ethos is much more aligned.

Something I"m definitely interested in having a go at once i"ve got my first boards.

Any other similar tools along these lines folk are interested in?

Andy

PiMSE
Posts: 5
Joined: Mon Nov 28, 2011 6:38 pm

Re: Statistics, Analytics and Modelling...

Sun Feb 05, 2012 6:23 am

I never heard of WPS before.  That looks like a neat idea.  And I see they've already ported to Linux...but not for free.  How complete is their SAS lanugage support?

Wikipedia's SAS Language page links to an article about a project called GNU DAP, which "was written to be a free replacement for SAS."  Hmm...the project homepage has a footer reading, "Updated: $Date: 2007/01/05".   And the latest entry on their FTP source download page is dated 2008 ... I'm not sure this is still an active project.

But, to answer your question:

I'd like to see R, and one of its GUI frontends (I've used R-commander) ported.

I just installed this pair under an Ubuntu variant on an Atom CPU netbook, just see if the Atom could handle it.  And it did.  I don't know how the ARM will do...

The R statistics language is growing in popularity right now, and broke into the TIOBE top 20: http://www.tiobe.com/index.php.....index.html

I don't know what's driving that growth, but maybe educators should pay some attention to it and help fill the demand.

I also liked Maxima.  Maxima is a "computer algebra system" and it has a WX-widgets GUI frontend called wxMaxima, that provides "pretty print" rendering of symbolic expressions, and plotting capability.  (A lot like Mathematica.)

I also installed Maxima and wxMaxima on my Atom system and they perform surprisingly well.  Again, I don't know how it will perform on ARM.

But, there's already a Debian package for Maxima on ARM, so that's encouraging.

The Gnumeric spreadsheet contains pretty good implementations of statistical functions, in contrast to MS Excel: http://www.csdassn.org/softwar.....eports.cfm

And there's RGnumeric that lets you call R functions from Gnumeric...but their changelog looks like it was most recently updated in 2003.  So, I'm not sure how current that is, but the mailing list still has some traffic, as of this month (Feb 2012):

http://mail.gnome.org/archives.....index.html

Both the R and Maxima projects have Debian packages.  I installed Maxima straight from the Ubuntu package repositories.  And, while the R packages had to be installed from a 3rd party repository, Ubuntu packages were available.

PiMSE
Posts: 5
Joined: Mon Nov 28, 2011 6:38 pm

Re: Statistics, Analytics and Modelling...

Sun Feb 05, 2012 6:38 am

For what it's worth, it looks like there are also Debian packages for R on arm:

http://packages.debian.org/sea.....-base-core

andyl
Posts: 265
Joined: Tue Jan 10, 2012 11:05 am

Re: Statistics, Analytics and Modelling...

Sun Feb 05, 2012 10:43 am

PiMSE said:


But, to answer your question:

I'd like to see R, and one of its GUI frontends (I've used R-commander) ported.


To get R working on a PI should be easy.  It will probably just require compiling on the Pi(or cross-compiling if someone has set up a Pi toolchain on a X86 based linux).

GNU PSPP (a SPSS clone) should also be a simple matter of an appropriate compile.

mdewey
Posts: 37
Joined: Wed Dec 07, 2011 10:47 am
Location: UK

Re: Statistics, Analytics and Modelling...

Sun Feb 05, 2012 11:48 am

AndyBolton said:


Any other similar tools along these lines folk are interested in?

Andy


Andy,

I run R satisfactorily on an EEEPC 701 which is not completely unlike the power of the RPi so you should have no problems with running it (as I intend to). Probably using Debian would be simplest as the repositories have it compiled for ARM. Whether it uses hardware floating point is something we may have to wait and see. If not we would need to compile it ourselves. Since most of the developers of R seem to use Linux and are very helpful all should go smoothly.

Of course then you have to learn R which takes a while but is well worth it.

kme
Posts: 448
Joined: Sun Sep 04, 2011 9:37 am

Re: Statistics, Analytics and Modelling...

Sun Feb 05, 2012 11:58 am

mdewey said:


AndyBolton said:


Any other similar tools along these lines folk are interested in?

Andy


Andy,

I run R satisfactorily on an EEEPC 701 which is not completely unlike the power of the RPi so you should have no problems with running it (as I intend to).


The CPU in eeePC 701 is 50% faster (measured in BogoMips) than the R-Pi and even more important I think it's Intel graphics is HW accelerated too (unlike VideoCore IV).

Edit: I talk about x.org acceleration. Obviously video playback is accelerated.

mdewey
Posts: 37
Joined: Wed Dec 07, 2011 10:47 am
Location: UK

Re: Statistics, Analytics and Modelling...

Sun Feb 05, 2012 2:05 pm

kme said:


mdewey said:


AndyBolton said:


Any other similar tools along these lines folk are interested in?

Andy


Andy,

I run R satisfactorily on an EEEPC 701 which is not completely unlike the power of the RPi so you should have no problems with running it (as I intend to).


The CPU in eeePC 701 is 50% faster (measured in BogoMips) than the R-Pi and even more important I think it"s Intel graphics is HW accelerated too (unlike VideoCore IV).

Edit: I talk about x.org acceleration. Obviously video playback is accelerated.



To be fair I did say '... not completely unlike ...' and in my view 50% is within that range. Mind you I am an academic and where lay people say 'There is a cow in the field' we say 'It looks not unlike a cow from this side.' I use R from the command line so the graphics speed is not usually an issue for me unless I am specifically drawing a graph.

jackb
Posts: 6
Joined: Mon Dec 05, 2011 11:59 pm

Re: Statistics, Analytics and Modelling...

Tue Feb 07, 2012 2:41 pm

You could install rstudio on the pi... but then again it might make more sense to run rstudio server on a more powerful machine and use the pi as a client.

Roti2000
Posts: 3
Joined: Thu Jul 26, 2012 3:51 pm

Re: Statistics, Analytics and Modelling...

Fri Jul 05, 2013 3:56 pm

I wanted to resurrect this thread and see if anyone has gotten R and/or RStudio up and running on a pi?

Return to “Staffroom, classroom and projects”