I thank Naomi Robbins, Madeline Bauer, Matt Austin, and Richard
Heiberger for their advice on my question, which was:
“I have data for four situations, and for each I have a few (c. 4--6)
original data points, and about 100 points for non-linear functions fit
to those data. (The functions differ among the situations.) Is there a
way to use trellis to make a 2x2 grid of plots, each containing the raw
data as points, and the fitted functions as lines?”
The answer that worked best for me was one that Richard sent off-list to
me, namely:
observed <- data.frame(y=rnorm(12),
x=1:12,
case=c(1,1,1,2,2,2,3,3,3,3,4,4),
form=1)
aa <- data.frame(y=rnorm(14), x=1:14, case=1, form=2)
bb <- data.frame(y=rnorm(16), x=1:16, case=2, form=2)
cc <- data.frame(y=rnorm(17), x=1:17, case=3, form=2)
dd <- data.frame(y=rnorm(19), x=1:19, case=4, form=2)
tmp <- rbind(observed, aa, bb, cc, dd)
tmp$case <- factor(tmp$case)
tmp$form <- factor(tmp$form, labels=c("raw", "fit"))
xyplot(y ~ x | case, pch=16,
panel=panel.superpose,
group=form, type=c("p","l"), data=tmp)
Madeline’s suggestion to use print.trellis would likely also have worked
well.
Thanks to all for your help.
Dave Parkhurst
|