[R-sig-Geo] How to obtain standard errors of local regression coefficients in spgwr::ggwr()?

[Roger Bivand]

Does ?spgwr::ggwr state that any such values are returned? It says:

"a SpatialPointsDataFrame (may be gridded) or SpatialPolygonsDataFrame 
object (see package "sp") with fit.points, weights, GWR coefficient 
estimates, dispersion if a "quasi"-family is used, and the residuals of 
type "type" in its "data" slot."

Nothing else is returned. You may like to try the GWmodel package, which 
may offer some further output, but this is a largely unexplored area, and 
in any case inference is not encouraged for GW output as the technique is 
exploratory.

Hope this clarifies.

Roger

PS. Please post plain text, not HTML-formatted.

On Sun, 6 Jun 2021, mr via R-sig-Geo wrote:

> Dear list,
>
> I am using spgwr::ggwr() to fit generalized geographically weighted regression with Poisson model and log-link function. The results provide local coefficient estimates, but i am missing how to get their standard errors (or t statistics) to compute pseudo p-values.
> Below is a toy example using SpatialEpi::NYleukemia dataset:
>
> # --------
>
> library(SpatialEpi)
> library(spgwr)
>
> ## Load data
> data(NYleukemia)
> population <- NYleukemia$data$population
> cases <- ceiling(NYleukemia$data$cases * 100)
> centroids <- latlong2grid(NYleukemia$geo[, 2:3])
>
> # data frame
> nyleuk <- data.frame(centroids, cases, population)
>
> # set coordinates as vector
> coordny <- cbind(centroids[,1],centroids[,2])
>
> # set a kernel bandwidth
> bw <- 0.5
>
> # fit ggwr()
> m_pois <- ggwr(cases ~ offset(log(population)),
>               data = nyleuk, gweight = gwr.Gauss,
>               adapt = bw, family = poisson(link="log"),
>               type="working", coords = coordny)
>
> # returns spatial point with coefficients
> # but no standard errors :(
> head(m_pois$SDF using data)
>
> # -------
>
> Is there any way to get standard errors of the local coefficients?
> Thank you, Manuel
>
>
> Sent from Mail for Windows 10
>
>
> 	[[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo using r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>

-- 
Roger Bivand
Emeritus Professor
Department of Economics, Norwegian School of Economics,
Postboks 3490 Ytre Sandviken, 5045 Bergen, Norway.
e-mail: Roger.Bivand using nhh.no
https://orcid.org/0000-0003-2392-6140
https://scholar.google.no/citations?user=AWeghB0AAAAJ&hl=en