My apologies for the duplicate list, but I forgot to use plain text in my last
post, and thus there was a problem posting to archive. Here is the post now:
Hi Everyone,
I'm new to R, so thank you in advance for your patience. I'm using US census
block groups in Buffalo New York to examine how neighborhood characteristics
affect crime. I would like to use an inverse distance weights (distance decay)
of block group centroids that is banded at 1 kilometer. In other words, I
would like to create a one kilometer buffer around each of the centroids, then
use row standardized inverse distance weights. Finally, I would like to run a
spatial error model using these weights.
Here is my code:
> library(foreign)
> library(spdep)
>
> buffalo <- read.dta("buffalo_blkgrps.dta")
> attach(buffalo)
> names(buffalo)
[1] "bgidfp00" "gage2900" "gavghhinc00" "gblack00" "gcrwd00"
[6] "gethhet00" "ghhincsdln00" "glatino00" "gocc00" "gowner00"
[11] "gpop00" "gpov00" "assaul" "robber" "burglr"
[16] "motveh" "murder" "larcen" "x" "y"
>
> #distance based neighbors - making a neighbor list - bounded to 1 kilometer
> coords <-(cbind(x,y))
> neigh.nb <- dnearneigh(coords, 0, 1, longlat=TRUE)
> summary(neigh.nb)
Neighbour list object:
Number of regions: 409
Number of nonzero links: 5298
Percentage nonzero weights: 3.167126
Average number of links: 12.95355
1 region with no links:
12
Link number distribution:
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
26
1 1 3 3 4 5 14 25 28 23 19 26 32 44 29 24 29 25 24 13 15 12 3 4 1 1
1
1 least connected region:
3 with 1 link
1 most connected region:
382 with 26 links
>
> #making inverse distance weights
> neigh.dist <- nbdists(neigh.nb, coords, longlat=TRUE)
> inverse <- lapply(neigh.dist, function(x) (1/(x^2)))
>
> #creating row standardized spatial weights from list
> neigh.listw <- nb2listw(inverse, style="W",zero.policy=TRUE)
Error in nb2listw(inverse, style = "W", zero.policy = TRUE) :
Not a neighbours list
> traceback()
2: stop("Not a neighbours list")
1: nb2listw(inverse, style = "W", zero.policy = TRUE)
>
>
> assault <- errorsarlm(assaul_r ~ gage2900 + gblack00 + gcrwd00 + gethhet00 +
> ghhincsdln00 + glatino00 + gocc00+ gowner00 + gpov00,
+ data=buffalo,listw=neigh.listw, zero.policy=TRUE)
Error in eval(expr, envir, enclos) : object 'assaul_r' not found
The model will run when I do not include the code with the inverse weights and
only interpoint distances, but it's unclear to me why I can't include the
inverse distances when using nb2listw. Any ideas on why this is occurring
and/or help to alleviate this issue would be much appreciated. The spatial
error model from errorsarlm (package=spdep) will also run fine when I don't
include the inverse weights. Is there a better way to go about running this
spatial error model using row standardized inverse distance weights? Would
spautolm from the spdep package be better suited for these data? Any comments
or suggestions are welcome.
Thank you for your time!
Adam
Adam Boessen
Doctoral Student
Department of Criminology, Law and Society
University of California, Irvine
aboes...@uci.edu
p.s. If it's helpful, here is my version of R:
> version
_
platform i386-apple-darwin9.8.0
arch i386
os darwin9.8.0
system i386, darwin9.8.0
status
major 2
minor 10.1
year 2009
month 12
day 14
svn rev 50720
language R
version.string R version 2.10.1 (2009-12-14)
_______________________________________________
R-sig-Geo mailing list
R-sig-Geo@stat.math.ethz.ch
https://stat.ethz.ch/mailman/listinfo/r-sig-geo
