Generation of Neighbourhood Files for Dorling Cartograms and Generalization Filtering

1. Using R

The following R program calculates neighbourhoods starting with a ArcInfo shapefile. There are some R packages required. Here the code:

#################################################################  
# INPUT               
# shapefileName: the name of the shapefile without ".shp"
# shapefileDirectory: the name of the directory of the shapefile
# ids: the ID variable in the shapefile defining the sorting order 
#      of the boundary file and the data file (case sensitive!)
# OUTPUT
# The resulting neighbour file has the name shapefileName + ".fil"

shapefileName = "Polygone_SH"
shapefileDirectory = "."
idName = "BFS2009"    

#################################################################                 

# library(ctv)
# install.views("Spatial")

library(spdep)
library(rgdal)

sh <- readOGR(".", shapefileName)
ids = sh[[idName]]  

nb <- poly2nb(sh)

plot(sh, border="grey")
plot(nb, coordinates(sh), add=TRUE)
title(main=paste("Nachbarn im Shapefile",shapefileName, sep="\n"))

cards <- card(nb)
maxconts <- which(cards == max(cards))
if(length(maxconts) > 1) maxconts <- maxconts[1]
fg <- rep("grey", length(cards))
fg[maxconts] <- "red"
fg[nb[[maxconts]]] <- "green"
plot(sh, col=fg)
title(main="Region with largest number of contiguities")

plot(sh, border="grey")
plot(nb, coordinates(sh), add=TRUE)
title(main=paste("Neighbours in Shapefile",shapefileName, sep="\n"))

dsts <- nbdists(nb, coordinates(sh))
idw <- lapply(dsts, function(x) 1/(x / 10))


nbWeight <- function(nb, style, showNr) {
  nbw <- nb2listw(nb, glist=idw, style=style, zero.policy=T)
  cat(style, summary(unlist(nbw$weights)), "\n")
  cat(style, summary(sapply(nbw$weights, sum)), " (Sum)\n")
  if (showNr >= 0) cat(style, unlist(nbw$weights[showNr]), " - area", showNr, "\n" )
  nbw
}

nbw <- nbWeight(nb, "W", 24)
nbw <- nbWeight(nb, "B", 24)
nbw <- nbWeight(nb, "C", 24)
nbw <- nbWeight(nb, "U", 24)
nbw <- nbWeight(nb, "S", 24)

expandNBleft <- function(x) {
  rrr <-  vector(mode="integer", length(unlist(nbw$neighbours)))
  k = 0
	for(i in 1:length(x)) {
		for(j in 1:length(x[i][[1]])) {
			k = k + 1
			rrr[k] = i
		}
	}
	rrr
}

w1 <- sort(ids)
w0 <- 1:length(w1)
w <- as.data.frame(cbind(w0,w1))

v1 <- expandNBleft(nbw$neighbours)
v2 <- unlist(nbw$neighbours)
v3 <- 100* unlist(nbw$weights)
nb.df <- as.data.frame(cbind(v1,v2,v3))

rr <- merge(nb.df, w, by.x = "v1", by.y = "w0")
rr <- merge(rr, w, by.x = "v2", by.y = "w0")
res <- as.data.frame(cbind(rr$w1.x, rr$w1.y,rr$v3))

summary(res)

options(scipen=12)
filFilename = paste(shapefileName, ".fil", sep="")
write.table(res, file=filFilename, col.names=F, row.names=F)

cat ("Resulting neighbour file:", filFilename, "\n")

#################################################################                 
                          
              

2. Using ArcGIS and Excel

(mainly for internal documentation)

Step 1: ArcInfo-AML (ArcInfo required)

shp2fil.aml

Input: ArcInfo-Shapefile with an ID-Item; output: 3 text files for the subsequent processing step.

Step 2: Excel-Macro (MS Excel required)

shp2fil.xls

The content of the 4 files from step 1 were pasted into the three sheets "aat", "pat", "reg" and "nam". The macro creates the .fil data in the sheet "fil".