R help - Apr 2005 - Howto overlay two plots and save them in one pdf file?

Hello,

I do have two different plots from LAD and OLS regression objects like 
this:

# LAD regression and related plot
rq(formula = SBP ~ Age)
f = coef(rq(SBP ~ Age))
pred = f[1] + f[2]*Age
plot (Age, SBP)
lines (Age, pred)

# OLS regression and related plot
Pred = lm(SBP ~ Age)
plot (Age, SBP)
lines (Age,fitted(Pred))



Well for comparatative reason I would would like to subsume both plots 
into a unifying plot and save them in one file.pdf.
I tried to find an answer in FAQ and mailinglists archive, no luck. 
Maybe did miss an appropiate answer to my question, so a pointer to 
solve my problem woud be sufficient!

Much thanks

Thomas


My system:


platform i686-pc-linux-gnu
arch     i686             
os       linux-gnu        
system   i686, linux-gnu  
status                    
major    2                
minor    0.1              
year     2004             
month    11               
day      15               
language R

> Well for comparatative reason I would would like to subsume both plots 
> into a unifying plot and save them in one file.pdf.
> I tried to find an answer in FAQ and mailinglists archive, no luck. 
> Maybe did miss an appropiate answer to my question, so a pointer to 
> solve my problem woud be sufficient!
Do you mean a second y axis? If so then something like this would do it:

library(quantreg)
Age <- rnorm(50)
SBP <- Age * runif(50)

# LAD regression and related plot
rq(formula = SBP ~ Age)
f = coef(rq(SBP ~ Age))
pred = f[1] + f[2]*Age
# leave room for the second axis
pdf("junk.pdf")
par(mar = c(3,3,1,3), mgp = c(2, 1, 0))
plot (Age, SBP, ylab = "SBP: QR Fit")
lines (Age, pred)
par(new=T)
# OLS regression and related plot
Pred = lm(SBP ~ Age)
plot (Age, SBP, col = "blue", ylab = "", xlab =
"", axes = F, pch = "+")
lines (Age,fitted(Pred), col = "blue")
axis(4, at=pretty(range(SBP)))
mtext("SBP: OLS Fit", 4, 2, col = "blue")
dev.off()

HTH, Andy

> > Do you mean a second y axis? If so then something like this would
> > do it
>
> Not exactly, I would also take advantage of overlaying the fitting
> curve of LAD on OLS, since both rely on the same dataset. Maybe two
> regression lines with varying shapes (i.E. straight versus dotted
> line)
Like this?

library(quantreg)
Age <- rnorm(50)
SBP <- Age * runif(50)

pdf("junk.pdf")
# plot the data
plot (Age, SBP, ylab = "SBP")
# LAD regression and related line
abline(rq(formula = SBP ~ Age), col = "red")
# OLS regression and related line
abline(lm(SBP ~ Age), col = "blue", lty = "dotted")
# make a legend
legend(-2,1, c("LAD", "LM"), col = c("red",
"blue"), text.col= c("red",
"blue"), lty = c("solid", "dotted"))
dev.off()

HTH, Andy