Hi everyone,
I'm trying to plot density curves on the axes of a plot, in a
similar way to the 'rug' function.
I have had a look at a few approaches and libraries, including:
* layout http://addictedtor.free.fr/graphiques/RGraphGallery.php?
graph=78
* fancygraph http://addictedtor.free.fr/graphiques/RGraphGallery.php?
graph=81
* ggplot - gave up since it needs R 2.3.0 and today I need to run on
less than that
* grid - probably powerful, good approach, but too much effort for a
midnight hack
* rug - since it does something similar
* lattice - it has multiple graphs, but I think they may need to be
homogeneous
* plot + lines
* plot + plot
And here is the solution I am using at the moment, based upon layout.
Can anyone suggest a better way to do this?
-Alex Brown
===# some data
x1 = rnorm(100)
y1 = rnorm(100)
# setup the layout
opar = par(no.readonly=TRUE)
omar = par("mar")
l = layout(matrix(c(2,0,1,3),2),c(1,4),c(4,1))
par(cex=1,bty="n")
# first plot : main plot
par(mar=omar * c(0,0,1,1)) # just top and right margins
plot(x1, y1, axes=FALSE, main="Density Margins",
xlab="", ylab=""
)
# vars
usepoly <- TRUE
plott = ifelse(usepoly,"n","l")
par(col="blue")
# second plot : y axis density
par(mar=omar * c(0,1,1,0))
par(xpd=NA)
yd = density(y1,from=min(y1),to=max(y1))
ydd= data.frame(x=-yd$y, y=yd$x)
plot(ydd, xaxt="n",xlim=c(0,min(ydd$x)),type=
plott,bty="n",ylab="y
(density)",xlab="")
if(usepoly) {
ydd=rbind(ydd, c(0,max(ydd$y)), c(0,min(ydd$y)))
polygon(ydd,col="grey",border=NA)
}
# third plot : x axis density
par(mar=omar * c(1,0,0,1))
par(xpd=NA)
xd = density(x1,from=min(y1),to=max(y1))
xdd= data.frame(x=xd$x, y=-xd$y)
plot(xdd, yaxt="n",ylim=c(0,min(xdd$y)),type= plott,xlab="x
(density)",ylab="")
if(usepoly) {
xdd=rbind(xdd,c(max(xdd$x),0), c(min(xdd$x),0))
polygon(xdd,col="grey",border=NA)
}
par(opar)
=== [[alternative HTML version deleted]]
