[R-sig-Geo] GWR Output Attributes

Roger Bivand Roger@B|v@nd @end|ng |rom nhh@no
Tue Mar 26 13:52:35 CET 2019


On Tue, 26 Mar 2019, Roger Bivand wrote:

> On Tue, 26 Mar 2019, James Garrett wrote:
>
>>  Dear R-Sig-Geo List,
>>
>>  I am hoping to verify the output variables from a gwr model (package
>>  spgwr). I've searched through the archives and haven't been able to find
>>  exactly what a few are, although I have a strong guess from the package
>>  description:
>>
>>  SDF a SpatialPointsDataFrame (may be gridded) or SpatialPolygonsDataFrame
>>  object (see package "sp") with *fit.points, weights, GWR coefficient
>>  estimates, Rsquared, and coefficient standard errors in its "data" slot.*
>
> There is no need to guess. Help pages have examples. Run:
>
> library(spgwr)
> example(gwr)
> str(col.gauss)
> str(col.gauss$SDF)
> names(col.gauss$SDF)
>
> to examine the contents of the objects of the first run. Note that the rest 
> of the example script shows why GWR is unreliable.

I've just used pkgdown to commit the processed help pages of spgwr to 
R-Forge:

http://rspatial.r-forge.r-project.org/spgwr/reference/gwr.html#examples

is the rendered version of running example(gwr) yourself, and is less 
effective because you can't play with the output.

Roger

>
> Don't guess, don't google or SO, just use the examples in the help pages 
> actively, for example changing the formula or argument values to see what the 
> arguments do. Do use alternative implementations to check your assumptions, 
> such as GWmodel. Do read vignettes: vignette("GWR").
>
> Hope this helps,
>
> Roger
>
>
>>
>>  Is this correct?
>>  sum_w = sum of the weights
>>  Intrc = X Intercept
>>  gtVI(F) = GWR coefficient estimates of the X intercept
>>  gwr_e = gwr residuals
>>  pred = Y prediction
>>  localR2 = local R2.
>>
>>  I apologize if this is rudimentary. Thanks so much in advance,
>>
>>  James
>> 
>> 
>
>

-- 
Roger Bivand
Department of Economics, Norwegian School of Economics,
Helleveien 30, N-5045 Bergen, Norway.
voice: +47 55 95 93 55; e-mail: Roger.Bivand using nhh.no
https://orcid.org/0000-0003-2392-6140
https://scholar.google.no/citations?user=AWeghB0AAAAJ&hl=en



More information about the R-sig-Geo mailing list