Problem nblag_cumul
On Thu, 28 Oct 2010, Roger Bivand wrote:
On Thu, 28 Oct 2010, Hauser, Christoph wrote:
Hi all! As a newcomer to both r and spatial econometrics I am trying to become familiar with both worlds. I have a question which is hopefully not too trivial: I am trying to create a second order queen matrix for European regions. As indicated in the documentation on spdep nblag and cumul_nblag are the appropriate commands to carry out this task (starting from the poly2nb command). Both commands worked fine, but they provided me with a second order matrix with one region having zero second order neighbors (IE01 - Central Ireland). When I tried to convert the nb object from cumul_nblag to a listw object the command failed with the error message that one region has zero neighbors. This struck me as odd given that the combination of first and second order matrices should make sure that each region has a neighbor (I don't have islands in my set). After inspecting the two neighbor matrices I noticed that the combination of the two matrices based on cumul_nblag worked out for every region except the one with zero second order neighbors. My question now is if it is customary for nblag_cumul to endow each region that has zero second order neighbors also in the cumulated matrix with zero regions? Does that make sense from a theoretical point of view?
Please put the input data and your code on a website, post the link, and I'll try to see whether this is intended or not. It may well be that the region with no second order neighbours is causing havoc.
Is the problem that you have a data set (maybe from GISCO) that includes different NUTS levels in the same file? The one I'm looking at includes IE, IE0 - identical, all Eire, then IE01 and IE02, which divide Eire into two parts at NUTS2, then IE01?, IE02?, in total 8 NUTS3 regions? Which level of NUTS are you working at, and where have your boundaries come from? Are you sure that your file does not include multiple levels - I can reproduce the problem when multiple levels are present, but not otherwise. Roger
Roger
I would be grateful for any help in this regard! Christoph PS: I am happy to provide my dataset if that is helpful. Dr. Christoph Hauser University of Innsbruck Department of Economic s Universitaetsstrasse 15 6020 Innsbruck Tel.: 0043 512 507 7354 Fax: 0043 512 507 2980 [[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