[R-sig-Geo] ArcGIS SWMs in R

[Roger Bivand]

On Mon, 26 Apr 2010, Scott L Minkoff wrote:

> To be a bit more precise than I was in the previous note, the full error
> message I get is:
>
> Error in res[i, listw$neighbours[[i]]] <- listw$weights[[i]] :
>  NAs are not allowed in subscripted assignments

That helps a little - the output of traceback() would have helped more, 
and shows (by grepping the source code) that the error occurs in 
listw2mat(), so at least one of listw$neighbours[[i]]] is an NA. Since I 
have a copy of the input data file, I see that:

> summary(unlist(lw$neighbours))
    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's
     2.0   336.0   668.0   667.9  1006.0  1330.0   137.0

If you form the sn object dropping the possibly spurious +1 in 
to=match(xx$NID+1, to_IDs):

sn <- data.frame(from=match(xx$FID2, from_IDs), to=match(xx$NID,
    to_IDs), weights=xx$WEIGHT)

the sn object converts to a listw object and on to a matrix without 
complaining. match() by default returns NA when no match is found. I think 
the original +1 came from being misled by the 0-base in the FID values.

Hope this helps,

Roger


>
> Thanks,
> Scott
>
> -----Original Message-----
> From: r-sig-geo-bounces at stat.math.ethz.ch
> [mailto:r-sig-geo-bounces at stat.math.ethz.ch] On Behalf Of Scott L Minkoff
> Sent: Monday, April 26, 2010 1:55 PM
> To: r-sig-geo at stat.math.ethz.ch
> Subject: [R-sig-Geo] ArcGIS SWMs in R
>
> Hello,
>
> I am still fairly new to spatial data analysis using R and would appreciate
> any help the list might be able to provide.  I have been using Roger's code
> to bring ArcGIS Spatial Weights Matrices into R (see below).  However, when
> I go to run spatial lag models, I get the following error:
>
> "NAs are not allowed in subscripted assignments"
>
> I have tried the various na.action arguments to no avail.  Note that the
> objects seem to work fine for Moran's I tests.  Does anybody know how to
> solve this problem?
>
> Thanks in advance,
> Scott
>
> Roger's Script for bringing Arc SWMs into R and making them list objects:
>
> library(foreign)
> x <- read.dbf("spatialweightmatrix3.dbf")
> from_IDs <- sort(unique(x$FID2))
> to_IDs <- sort(unique(x$NID))
> # find the IDs used
> all.equal(from_IDs, to_IDs)
> o <- order(x$FID2, x$NID)
> # order the object
> xx <- x[o,]
> # convert manually to a spatial.neighbours representation
> sn <- data.frame(from=match(xx$FID2, from_IDs), to=match(xx$NID+1,
>   to_IDs), weihts=xx$WEIGHT)
> attr(sn, "n") <- length(from_IDs)
> attr(sn, "region.id") <- as.character(from_IDs)
> class(sn) <- c("spatial.neighbour", class(sn))
> library(spdep)
> lw <- sn2listw(sn)
> lw
> summary(sapply(lw$weights, sum))
> table(card(lw$neighbours))
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>
> _______________________________________________
> 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