[GRASSLIST:5815] Re: linear regression of 2 rasters

[Roger Bivand]

On Wed, 16 Feb 2005, Dylan Beaudette wrote:

> Hi everyone,
> 
> Is there any simple way in GRASS to do a linear regression of 2 raster files, 
> or would I need to first subsample with vector points?
> 
> Or- is there anyway to do this in gnuplot / R.
> 
> I have found that R is not behaving like it used to, when used with GRASS54...

See the STATGRASS mailing list. The switch to the 3D WINDOW means that 
regular service is currently only supported for GRASS up to and including 
5.4. The work-around is to use interp=TRUE in gmeta() and perhaps 
elsewhere, because the compiled code in the R/GRASS interface only works 
with 2D WINDOWs, and there is nothing to condition against in 6.0.0beta 
(there will be an environment variable in 6.0.0 when it comes). If the 
interpreted interface is good enough (exchanging ASCII files), 6.0.0beta 
may work.

Once 6.0.0 stabilises, a fresh version of the interface will appear. If 
people want to help, I'll put it on sourceforge and give developer write 
access if that would help. It isn't obvious how to do this right, because 
it is a contributed package to both R and GRASS, so has to suit build 
trains and constraints across platforms, and keep working for 5.* users.

Roger

> 
> Any ideas?
> 

-- 
Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Breiviksveien 40, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 93 93
e-mail: Roger.Bivand at nhh.no