[R-sig-Geo] "txt" to "listw"

[Roger Bivand]

On Wed, 21 Mar 2012, anny peng wrote:

> Dear R users,
>
> Thank Dr. Divand for creating spdep, and thank everyone for your attention.
>
> I attach a 30*30 matrix, which consists of 0 and 1.  The matrix is  a
> "txt" file.  I would like to change this matrix into a ".listw" file
> so that I can estimate spatial auto-correlation.
>
> As a new user, I have difficulties in reading "txt" file and use the
> function "mat2listw".
>
> Below are my errors.
>
>
> data1= read.fwf("30_30.txt", widths=1)

The problem was resolved off-list. The main issue was about reading the 
tab-separated file into a matrix for use in mat2listw():

mat0 <- read.table("30_30.txt", header=FALSE, sep="\t", fill=TRUE)

and converted into a matrix with:

mat <- as.matrix(mat0)
dim(mat)

However, it has NAs, not 0, where there are no neighbours, so this must be
repaired:

summary(c(mat))
mat[is.na(mat)] <- 0
summary(c(mat))

Reading the R Data Import/Export manual would have helped, together with 
the use of CSV as the spreadsheet output format - the fields here are not 
of fixed width, although they look like that; tabs are used as field 
separators, but fields with 1s in have a width of 2, while empty fields 
have a width of 1.

Finally:

lw <- mat2listw(mat, style="B")
summary(lw)

Roger

>
>
> coords <- coordinates(data1)
> col005 <- dnearneigh(points, 0, 1, attr(col.gal.nb, row.names = IDs))
> summary(col005)
> col005.w.mat <- nb2mat(col005, zero.policy=TRUE)
> col005.w.b <- mat2listw(col005.w.mat)
>
> ......
>
>
>
> Best regard
> Zz
>

-- 
Roger Bivand
Department of Economics, NHH Norwegian School of Economics,
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