[GRASS-stats] Any recent code changes in grass' raster engine or spgrass6

Nikos Alexandris nikos.alexandris at felis.uni-freiburg.de
Wed Mar 10 10:32:10 EST 2010


Hi devs!


I am using i.pca again and I've noticed some differences in the
resulting numbers (eigen-values, vectors and importance). After browsing
the source code, I found that there was a bug-fix 3 months ago [1]. This
explains some previous disagreement between grass' i.pca and R's
function(s) prcomp (for example).

That's good news, Thanks :-)

However, I am still troubled as to why R's prcomp() function gives
different results than it did some months before. I run the scripts I
used before on the same unchanged data and prcomp() gives completely
different numbers.

I could not find any significant code-change(s), while searching the
respective changelog(s), for prcomp() nor for spgrass6. I don't want to
trouble you with my problems but I am forced to ask:

was there recently any _very_ important change in the raster
engine (grass64, grass65), the region settings perhaps, or the way
grass-rasters are imported in R via spgrass6 in the last months?

Apologies for cross-posting,
Nikos

---
[1]
http://trac.osgeo.org/grass/changeset/39800/grass/branches/develbranch_6/imagery/i.pca/main.c




More information about the grass-stats mailing list