spdep (version 0.1-9)

BH: Belo Horizont 1994 homicide data

Description

The BH data frame has 81 rows and 4 columns for 81 planning units (UP) in the city. The dataset also includes BH.polys UP polygon boundaries, BH.xy UP polygon centroids, BH.bbs UP polygon bounding boxes, and BH.nb the neighbours list used in Assun��o and Reis (1999),

Usage

data(BH)

source

Data obtained from paper authors directly.

Details

The dataset was originally used to motivate and test a new proposal for adjusting Moran's I when the variable being tested for spatial autocorrelation is a rate (see EBImoran.mc()). The examples do not replicate the printed results because of an unfortunate re-ordering of the data - see the moran.test() with the re-ordered rate p to retrieve a printed result), but the substantive conclusions are sustained.

References

Assun��o RM, Reis EA 1999 A new proposal to adjust Moran's I for population density. Statistics in Medicine 18, pp. 2147--2162

Examples

data(BH)
plotpolys(BH.polys, BH.bbs)
plot(BH.nb, BH.xy, add=TRUE)
p <- BH$HOMICIDES / BH$POP
names(p) <- rownames(BH)
p.f <- ordered(cut(p*100000, breaks=c(0,1,6,15,40,625),
 include.lowest=TRUE))
names(p.f) <- names(p)
plotpolys(BH.polys, BH.bbs,
 col=rev(c("black", heat.colors(4)))[codes(p.f)])
legend(c(618000, 624000), c(781000, 787000),
 legend=c("0", levels(p.f)[2:4], "625"),
 fill=rev(c("black", heat.colors(4))))
title("Homicide rates per 100,000, Belo Horizonte, 1994")
joincount.mc(p.f, nb2listw(BH.nb, style="B"), nsim=999)
moran.test(p, nb2listw(BH.nb, style="B"), randomisation=FALSE)
tmp <- try(moran.test(p[order(BH$UP)], nb2listw(BH.nb, style="B"),
 randomisation=FALSE, spChk=FALSE))
print(tmp)
tmp <- try(moran.test(p[order(BH$UP)], nb2listw(BH.nb, style="B"),
 randomisation=FALSE, spChk=TRUE))
print(tmp)
moran.test(p, nb2listw(BH.nb, style="B"))
moran.mc(p, nb2listw(BH.nb, style="B"), nsim=999)
EBImoran.mc(spNamedVec("HOMICIDES", BH), spNamedVec("POP", BH),
 nb2listw(BH.nb, style="B"), nsim=999)
outlier <- which(p*100000 == 625)
BH[outlier,]
plot(log10(BH$POP), BH$HOMICIDES,
 pch=ifelse(p*100000 == 625, 1, 18), xaxt="n",
 xlab="UP population", ylab="Homicides",
 main="Homicides, Belo Horizonte, 1994")
axis(1, at=c(2,3,4), labels=c(100,1000, 10000))
text(log10(BH$POP)[outlier], BH$HOMICIDES[outlier],
 pos=1, adj=0.5, paste("UP", BH$UP[outlier]))
moran.mc(p[-outlier],
 nb2listw(subset(BH.nb, !(attr(BH.nb, "region.id") == BH$UP[outlier])),
 style="B"), nsim=999)
EBImoran.mc(spNamedVec("HOMICIDES", BH)[-outlier],
 spNamedVec("POP", BH)[-outlier], nb2listw(subset(BH.nb,
 !(attr(BH.nb, "region.id") == BH$UP[outlier])), style="B"), nsim=999)