Lattice and horizontally stacked density plots

Hello,

I am try to make a density plot where plots are stacked like the one
found here:

http://dsarkar.fhcrc.org/lattice/book/images/Figure_14_03_stdBW.png

I am facing problems, however. Using the code example below, I'd like
to generate a separate panel for each val of id2. Within each panel,
I'd like to have individual histograms each on separate lines based on
the value of id1. ?Note that the code example works fine if I use
"boxplot" instead of "densityplot". Any pointers would be much
appreciated.


library(lattice)
val<-rep(rnorm(10),100)
id1<-sample(c(1:5), 100, replace = TRUE)
id2<-rep(c(6:10),100, replace = TRUE)
a_df<-data.frame(cbind(id1, id2, val))
densityplot(factor(id1) ~ val | factor(id2), data=a_df)

-Manish

With lattice alone you can easily get all curves on the same level:
densityplot(~ val | factor(id2), groups=factor(id1),data=a_df,pch='|')

But if that doesn't do it for you, you could write your own panel
function. I don't have time to try it but I'm thinking one of these
might work
1)
create new ylim[1] from current.panel.limits()$ylim / number of groups
dens<- density(x)
use lines and polygons to draw the curves dens$x,dens$y at each new ylim level.

2) grid may come in handy here, splitting each panel into several viewports ?

3) use bwplot for "setup" but plot polygons of density (steps 2:3 from option 1)

Good luck with that.

Elai



On Wed, Feb 22, 2012 at 3:01 PM, Manish Nag <nagm01 at gmail.com> wrote:

Hello,

I am try to make a density plot where plots are stacked like the one
found here:

http://dsarkar.fhcrc.org/lattice/book/images/Figure_14_03_stdBW.png

I am facing problems, however. Using the code example below, I'd like
to generate a separate panel for each val of id2. Within each panel,
I'd like to have individual histograms each on separate lines based on
the value of id1. ?Note that the code example works fine if I use
"boxplot" instead of "densityplot". Any pointers would be much
appreciated.


library(lattice)
val<-rep(rnorm(10),100)
id1<-sample(c(1:5), 100, replace = TRUE)
id2<-rep(c(6:10),100, replace = TRUE)
a_df<-data.frame(cbind(id1, id2, val))
densityplot(factor(id1) ~ val | factor(id2), data=a_df)

-Manish