[R] points rejected as lying outside the specified window

[Rolf Turner]

This is a question about the "spatstat" package and as such would have
been more appropriately directed to the R-sig-Geo list.

(1) Your code seems unnecessarily complicated; I haven't followed it
through in detail however.

(2) You did not get an "error", you got a "warning".  This warning means
what it says:   A number of the points that you are trying to make into
a planar point pattern lie out side of the window that you specified.

Since you say that *all* of your points lie outside the specified window
one might guess that the coordinate system of the points in "myfile.csv"
(God how I hate that "mythis" and "mythat" terminology that Micro$soft
has introduced!) is different from the coordinate system that is used in the
"NYBorough" data set.

Look at

     range(coords[,1])
     range(coords[,2])
     range(myfile$X)
     range(myfile$Y)

to get some insight into the nature of the coordinates.

If the coordinates seem to be at least roughly commensurate
your could try:

WR <- ripras(x=myfile$X[u],y=myfile$Y[u],shape="rectangle")
Type_0 <- ppp(myfile$X[u],myfile$Y[u],window=WR)
WP <- owin(poly=list(x=coords[,1],y=coords[,2]))
plot(Type_0)
plot(WP,add=TRUE)

which might be revealing.

Other than that it's difficult to say what your problem is without having
access to your actual data.

     cheers,

         Rolf Turner

On 01/30/2013 09:29 AM, Abby Rudolph wrote:
> Hello,
>
> I am using the following code to create ppp files from csv data and map shape files, but I am getting some errors which I have been unable to fix by searching them online:
>
>
> library(spatstat)
> library(maps)
> library(maptools)
>
> NYC2<-readShapePoly("nybb.shp") # this is a map of the NYC boroughs without waterways and no census tract divisions (but it does include lines separating the 5 boroughs)
> plot(NYC2)
>
> NYBorough<-readShapePoly("NYBoroughsShapeMerge.shp") # this is a map of the NYC boroughs with census tract divisions
> plot(NYBorough)
>
> myfile<-read.csv("myfile.csv")
> # my csv file contains 4 columns (ID, variable_type, X, Y)
> # my goal is to determine whether individuals of variable_type=1 are spatially distributed more than what would be expected due to chance, if individuals of variable_type=0 are spatially distributed more than what would be expected due to chance, and finally if the spatial distribution of variable_type=1 differs from that of variable_type=0.
>
> x<-NYBorough at polygons[[1]]@Polygons
> # ppp(x.coordinates, y.coordinates, x.range, y.range)
> coords<-slot(x[[1]],"coords")
> coords
> coords<-coords[19:1,]
> coords<-coords[-1,]
> coords
>
> u<-myfile$type==0
>
> Type_0<-ppp(myfile$X[u],myfile$Y[u],
>                  window=owin(poly=list(x=coords[,1],y=coords[,2])))
> Type_1<-ppp(myfile$X[!u],myfile$Y[!u],
>                  window=owin(poly=list(x=coords[,1],y=coords[,2])))
>
> ERROR Message:
> Warning message:
> In ppp(myfile$X[u], myfile$Y[u], window = owin(poly = list(x = coords[,  :
>    217 points were rejected as lying outside the specified window
> Warning message:
> In ppp(myfile$X[!u], myfile$Y[!u], window = owin(poly = list(x = coords[,  :
>    435 points were rejected as lying outside the specified window
>
> # I only have 652 points, so all of my points are rejected as lying outside the specified window
>
> Can someone please tell me what I am doing wrong.  The code I was using as a template was only looking at one county so it had a defined border which was somewhat rectangular.  Could it be because the shapefile I am using for NYC has too many borders?  Is it the coords code that is incorrect?
>
> Any advice or suggestions would be greatly appreciated.
>
> In case it is helpful, the code that follows is:
> qtype_1<-quadratcount(type_1,10,10)
> plot(type_1,pch="+",main=Title")
> plot(type_1,add=TRUE,col="red",ltw=2,cex=1.2)
> polygon(coords,lwd=2)
> points(type_1,pch=".")