[GRASSLIST:4864] Re: Moran's I

[Roger Bivand]

Alexandru Codilean wrote:

> Dear All,
>
> I am experienced with GIS but new to GRASS and LINUX. I have installed
> GRASS 5.0.0 on my machine using the precompiled binaries for Linux (I am
> using RedHat 8.0) I cannot use the command r.moran. Grass cannot find
> it. I have looked myself and there is nothing on the machine with this
> name o any similar one. The documentation and command list includes this
> command for grass 5.0.0.

In fact the manual page is still distributed with the 5.0.0 source, but no
r.moran code is present. The code was written by Jo Wood, and the manual
page is dated 1994. My guess would be that the code has not been
maintained for some time, is maybe in the 4.2.1 source tarball, may be in
the 4.3 tarball, but hasn't migrated to 5.0.0. This is because the change
to 5.0.0 involved the replacement of numeric 0 by a NULL not available
value (since numeric 0 can be a useful numerical value), and migration to
floating point rasters for data of the kind returned by a local Moran's
I_i. If you need Moran's I, one posibility is the interface to R described
on the geostatistics section of the website. In R, Moran's I and local
Moran's I_i are in contributed package spdep. If you need specific help,
contact me off the list, and we can work something out (doing tests on
local Moran's I_i is not so easy - there is a saddlepoint approximation
function in spdep).

Roger

> What is wrong here? Could that be that the binary is missing this and
> possibly other files?
>
> Thank you!
>
> ____________________________________________________
>
>  Tiberiu Alexandru CODILEAN
>  SSA GIS Consultant
>  United Nations,  HQ, Cartographic Section, Room L-172
>  New York, NY 10017, U.S.A.
>  Tel.: +1 917 367 3274
>  Fax: +1 212 963 3742
>  E-mail: codilean at un.org


-- 
Roger Bivand
NHH, Breiviksveien 40, N-5045 Bergen, Norway
(travelling but still accessible)