[R-sig-Geo] distance problem in GWR

[Roger Bivand]

On Mon, 1 Feb 2010, Servet Ahmet Cizmeli wrote:

> Hello all,
>
> I am about to perform a GWR analysis using measurements collected at sea
> onboard a ship. Some of the stations are located at opposite sides of an
> island or within adjacent bays... Measuring bird's flight distances between
> the stations do not make sense at all those cases.
>
> Is there any way to presently instruct the R's implementation of GWR that
> the distances are to be measured along sea-ways? This would probably
> necessitate the concurrent input of a land mask (or polygon).
>
> As I presume the answer of the above question is NO (no mention of it
> anywhere in the GWR documentation), I would like to know how difficult
> would it be to add that functionality for a r-novice like me.

Look at the fish swims thread on this list recently:

https://stat.ethz.ch/pipermail/r-sig-geo/2010-January/007472.html

You'll see that several participants suggested using MDS to convert the 
non-Euclidic distance matrix to a 2-column coordinate matrix 
"representing" the relative position of the observations. You could use 
those with GWR, I think. It wouldn't be exact, but at least closer to your 
reality than bird flies.

Roger

>
> thanks
> Servet
>
> _______________________________________________
> 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