[R-sig-Geo] Can spgwr give the significance test of non-stationarity

[Roger Bivand]

On Fri, 11 Apr 2008, Shaofei Chen wrote:

> Dear All,
>
> Fothingham's "GWR" book chapter 9.2 gives the significance test of 
> non-stationarity using a Monte Carlo test and a Leung test. This is also 
> implemented in his software. I wonder can spgwr also give such tests? I 
> cannot find them in the outcome. Here is an example code

R functions are typically modularised, and only provide the components 
needed. The tests require the hat matrix. So if you want tests, ask for 
the hat matrix, and run the tests on that object - they are separate 
functions - see help(package=spgwr). With regard to the tests, mileage 
varies a lot, even within the book itself. Really, there are no degrees of 
freedom left for tests, are there? The technique was intended to be 
exploratory.

>
> library(spgwr)
> data1<-read.csv(......)
> bw <- gwr.sel(y1~ X1 + X2+ X3 , data=data1, coords=cbind(data1$X, data1$Y))
> data1.gauss <- gwr(y1 ~ X1+ X2 + X3, data=data1, coords=cbind(data1$X, data1$Y), bandwidth=bw, hatmatrix=TRUE)
> names(data1.gauss)
>
> it returns "SDF"       "lhat"      "lm"        "results"   "bandwidth" "adapt" "hatmatrix" "gweight"   "this.call"
>
> If I find one exogenous variable is not significant non-stationary, how 
> can I run the mixed GWR model (including varied coefficients and 
> constant coefficients)? Thank you in advance.
>

No mixed version is provided. I guess you could use an offset of the 
product of the variable and global coefficient if you really wanted to. 
But as mentioned above, GWR is not a model, it is a way of exploring 
missing variables and/or inappropriate functional forms.

Hope this helps,

Roger

> Best,
> Shaofei
> 	[[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>

-- 
Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Helleveien 30, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no