[R-sig-Geo] values of Moran´s I index

[Roger Bivand]

On Wed, 6 Apr 2011, Robert Pazur wrote:

> Dear all,
>
> with usage of spdep package I tried to explore the impact of changing
> location of occurrence of  1?s in binary fashion on Moran index  by shifting
> an artificial 9pixel- square (small square  left below in the picture) in
> all direction (based on 1 pixel move) in my squared area (500x500 pixels).
> www.scandinavia.sk/data/morans.jpg
> I was looking at the distances where the values of Moran?s I index decrease
> below particular threshold (in this case 0.6,0.5,0.4) and record this
> distances as some indexes in centroids of corresponding square.
> I expected that this values in my  area would by symmetrical in all
> directions.  But there are not.
> Have somebody some ideas why?

I'm afraid that there is far too little information here to know what your 
question is. Provide very concise sample code (for a 50 by 50 grid, for 
example) instead. You have not said how the weights were defined, or 
whether you are row-standardising the weights. The images suggest edge 
effects, but without a reproducible example, it isn't possible to say. Can 
you reproduce the same effect with a grid on a torus?

Roger

> Thanks in advance.
>
> Robert
> -------------------------------------------------------
> Robert Pazur
> PhD student
> Institute of Geography
> Slovak Academy Of Sciences
>
> 	[[alternative HTML version deleted]]
>
>

-- 
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