Message-ID: <Pine.LNX.4.44.0408161833200.878-100000@reclus.nhh.no>
Date: 2004-08-16T16:48:03Z
From: Roger Bivand
Subject: CAR spatial model functions?
In-Reply-To: <5.2.1.1.2.20040816111205.02f69600@gwia.pen.wisc.edu>
On Mon, 16 Aug 2004, Brian Aukema wrote:
> At 15 Aug 2004 14:43:35 +0200 (CEST), Roger Bivand <Roger.Bivand at nhh.no> wrote:
>
> > There was a question (as the original questioner noted) on the list some
> months ago
> > about CAR by ML (essentially doing what S-Plus SpatialStats does)? Is
> this a general
> > need? Should spdep do it (I've kept away from it because spdep was not
> intended to
> > duplicate things already available in some implementation of S, rather
> to complement)?
> > Has anybody any code that could be polished?
>
> Thanks to the list for the suggestions to implement CAR models with
> WinBUGS. I would
> love to see spdep include implementations of CAR via ML in the future, even
> though S-Plus
> has similar capabilities. R has a much better price, especially when being
> used as a
> teaching tool.
Does anyone have a well-documented data set and weights matrix for which
authoritative results of CAR estimation exist - preferably for a single
spatial parameter - that they would be willing to make available for
trying this out?
Roger
>
> Brian Aukema
>
>
> Brian Aukema, PhD
> Visiting Fellow, Canadian For. Service Pacific Forestry Centre
> 506 West Burnside Road
> Victoria, British Columbia
> V8Z 1M5
>
>
>
--
Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Breiviksveien 40, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 93 93
e-mail: Roger.Bivand at nhh.no