Skip to content

Varying grid.rect in different panels of a Lattice plot

2 messages · Leon Barmuta, Deepayan Sarkar

Original
Leon Barmuta
# 
Dear r-help,

Sleep-deprivation from having 2 youngsters under 2 around the house is 
fuzzing my brain, so please be gentle if the answer to this query is obvious!

In the example below, I'm trying to use grid.rect to add grey rectangles to 
the panels of a lattice plot to indicate which months spawning occurred of 
a (very cute) native Tasmanian fish. The fish in the two lakes spawned at 
slightly different times, so grid.rect needs to be conditioned on when 
spawning happened in each lake. However, the panel function I wrote first, 
and reproduce below, inserts grey rectangles for all dates that spawning 
occured pooled across the two lakes. So after dreging r-help, I've messed 
with subscripts and tried fiddling with panel.number, but can't get this to 
work. Any suggestions most welcome!

# Make up a short data set

library(lattice)
library(grid)

Lake <- rep(c("Crescent","Sorell"), each=13)
Spawning <- c("Y", rep("N",7), rep("Y",4), rep("N",8), rep("Y",5), "N")
Catch <- rpois(26, 30) # fake data
Plotdate <- rep(seq(as.Date("2000/10/1"), by="month", length=13),2)

trellis.device(theme=col.whitebg())

# panel function that doesn't quite work

myPanel <- function(x, y, ...) {
  grid.rect(x=unit(Plotdate[Spawning == "Y"], "native"),
       just="left", width=unit(31, "native"), # a bit of a fudge
       gp=gpar(col="transparent", fill="light grey"))
       panel.xyplot(x, y, ...)}

xyplot(Catch ~ Plotdate|Lake, type="b",
       panel=myPanel,
       layout=c(1,2), xlab="",
       scales = list(x=list(rot = 45, at=as.numeric(Plotdate),
                     labels=format(Plotdate, "%b%Y"))))

Regards,

Leon

-------------------------
Dr Leon A. Barmuta, Senior Lecturer in Zoology
School of Zoology & TAFI, University of Tasmania, Private Bag 5, Hobart, 
Tasmania 7001, Australia
Phone (03) 6226 2785;  Fax (03) 6226 2745; International callers replace 
(03) with +61 3
School of Zoology web page: http://www.scieng.utas.edu.au/zoo/
My web page: http://www.scieng.utas.edu.au/zoo/pagedetails.asp?lpersonId=222
Deepayan Sarkar
#
On Wednesday 16 March 2005 18:48, Leon Barmuta wrote:
If I understand you right, you could do something like:



myPanel <- function(x, y, groups, subscripts, ...)
{
    grid.rect(x = unit(x[groups[subscripts] == "Y"], "native"),
              just="left", width=unit(31, "native"), # a bit of a fudge
              gp = gpar(col="transparent", fill="light grey"))
    panel.xyplot(x, y, ...)
}


xyplot(Catch ~ Plotdate | Lake, type="b",
       groups = Spawning,
       panel = myPanel,
       layout= c(1,2), xlab="",
       scales = list(x=list(rot = 45, at=as.numeric(Plotdate),
                     labels=format(Plotdate, "%b%Y"))))



There's nothing (much) special about 'groups', you could also do 


myPanel <- function(x, y, flag, subscripts, ...)
{
    grid.rect(x = unit(x[flag[subscripts] == "Y"], "native"),
              just="left", width=unit(31, "native"), # a bit of a fudge
              gp = gpar(col="transparent", fill="light grey"))
    panel.xyplot(x, y, ...)
}


xyplot(Catch ~ Plotdate | Lake, type="b",
       flag = Spawning,
       panel = myPanel,
       layout= c(1,2), xlab="",
       scales = list(x=list(rot = 45, at=as.numeric(Plotdate),
                     labels=format(Plotdate, "%b%Y"))))



except that if you had your variables in a data frame (supplied as the 
'data' argument) and not in your search path, anything supplied as 
'groups' would also be evaluated in the data frame.

Deepayan