[R-sig-Geo] spgwr package alpha (depends on sp) on sourceforge
Roger.Bivand at nhh.no
Thu Jun 10 09:34:47 CEST 2004
Partly as a response to the discussion between Hisaji Ono and Danlin Yu,
and partly because Danlin has made lots of helpful contributions of ideas
and code, source packages and Windows binary packages for sp and spgwr are
now available on:
spgwr implements some of the methods used in Geographically Weighted
Regression (Stewart Fotheringham, Chris Brunsdon, Martin Charlton, 2002,
Wiley), but the implementations are not authoritative, and are intended
for exploring the potential of the proposed methods.
Remember to install sp first - spgwr depends on it!
Documentation is incomplete, please look at the examples or in function
code for further information; this will improve. Anybody have nice
examples that we can use?
The sp release notes are: Minor modifications to package - please note
this is still alpha, and any production work objects must also be stored
in externally readable formats, because the slots in the classes can
The spgwr notes: This very early alpha package builds on earlier work and
major contributions by Danlin Yu, and provides both GWR functions, the
Georgia data set from the GWR bo ok, and is at the same time a testbed for
using the S4 classes from the sp packa ge in an analytical setting. As
mentioned, this is a trial release, and both obj ect formats and the way
results are calculated will change - store all important data in
externally retrievable formats!
Please pass comments to me as package maintainer, input is very welcome!
On Wed, 19 May 2004, Hisaji ONO wrote:
> Thank you very much, Mr. Yu.
> It works correctly.
> > Just an add-on. Since the warning message indicates that package 'mva'
> > and 'ctest' have been merged into 'stats', when you are altering the
> > source code, you can also replace all 'mva' and 'ctest' with 'stats'.
> > I think the error message actually just indicates that the function
> > print.coefmat() that has been called in the gwr.whole.test() is no longer
> > part of R (you can find this infomation by using ?Defunct in R).
> > From the ?Defunct you can also find that the print.coefmat() function
> > is an older name for 'printCoefmat' with a different default for
> > 'na.print'. I therefore replaced all the print.coefmat in the source code
> > of spgwr with printCoefmat, and the program gives the correct responses.
> R-sig-Geo mailing list
> R-sig-Geo at stat.math.ethz.ch
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
More information about the R-sig-Geo