subset grid and neighbour list

Dear listers
This may be obvious however I have thrown in the towel after numerous futile attempts. Here is my problem. I have a grid 
with 5 levels including NA. I would like to perform join count stats on it. Prof. Bivand advised me to either add NA as 
a factor, or subset the grid and eliminate the NA's (I feel the later makes more sense)

I think the steps I need to take are

1. Drop the NA's in mygrid (and save the ids of the cells dropped)

2. subset the neighbors list using information from step1

Is there anyone familiar with this especially step 1

library(spdep)
set.seed(20060110)
input <- data.frame(expand.grid(x=1:10, y=1:10), z=rpois(100, 3))
coordinates(input) <- c("x", "y")
table(input$z)
is.na(input$z) <- input$z < 1 | input$z > 5
table(input$z)
input$z <- as.factor(input$z)
rook <- dnearneigh(coordinates(input), 0, 1)
rook
plot(rook, coordinates(input))
joincount.multi(input$z, nb2listw(rook, 
style="B"))
# Error in joincount.multi(input$z, nb2listw(rook, style = "B")) : 
#        NA in factor
keep <- !is.na(input$z)
rook_sub <- subset(rook, subset=keep)
input_sub <- subset(input, subset=keep)
rook_sub
# look out for no-neighbour cells - if present, use zero.policy=TRUE below
plot(rook_sub, coordinates(input_sub))
joincount.multi(input_sub$z, nb2listw(rook_sub, style="B"))

This is for a full grid, but if you have SpatialPixels or SpatialPolygons 
the order of operations will be the same, I think. You need to watch out 
for no-neighbour cells after subsetting, and use zero.policy=TRUE in both 
joincount.multi() and nb2listw() if they have been introduced.

Roger

Ronnie

_______________________________________________
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