Skip to content

plotting fit from ols( rcs() )

5 messages · David Winsemius, x, Rolf Turner

x
#
I have: f <- ols( y1 ~ (rcs(x1,3) + rcs(x2,3) + rcs(x1,3) %ia% rcs(x2,3) ) )

Then I do: plot(f)

This plots y1 against x2 for a given value of x1.

Is there a way to make it *also* plot y1 against x1 for a given value of x2?

If not, I guess I can do coplot(y1 ~ x1 | x2).

Thanks in advance,
sp
#
On May 2, 2009, at 4:09 PM, x wrote:

            
Are you using the Design/Hmisc packages? If so, then you could get  
informative plots with:

plot(f, x1=NA, x2=NA, method ="contour")  # assuming you have used the  
datadist facilities.

This should give you a two dimension display of what might be called  
iso-levels of y. method= "persp" also works but I find it gives you a  
plot that has shape but not as much quantitative content.

If you want the somewhat more boring plot that you asked for, and  
x2.spec is set as the level of x2 that you specify, then consider:

plot(f, x1=NA, x2=x2.spec)
3 days later
x
#
Hi,

While xYplot(...) below produces an empty pdf file, plot(...) works fine. The same xYplot(...) produces correct output if tried directly in R console. Any suggestions?

Thanks,
sp

# empty pdf
trellis.device("pdf", file="./resid1_lat.pdf");
xYplot(resid(f) ~ fitted(f), method='quantile', nx=20,
       ylim=c(-100,250), xlim = c(-25,150),
       abline=list(h=0, lwd=0.5, lty=2),
       xlab="Fitted Values [lat]",
       ylab="Residuals [lat]")
dev.off();

# non-empty pdf
trellis.device("pdf", file="./resid1_lat.pdf");
plot(resid(f) ~ fitted(f), method='quantile', nx=20,
       ylim=c(-100,250), xlim = c(-25,150),
       abline=list(h=0, lwd=0.5, lty=2),
       xlab="Fitted Values [lat]",
       ylab="Residuals [lat]");
dev.off();
#
On 6/05/2009, at 12:50 PM, x wrote:

            
RTFFAQ (7.22)

	cheers,

		Rolf Turner

######################################################################
Attention:\ This e-mail message is privileged and confid...{{dropped:9}}
x
#
print() fixed it. Thank you!

sp
--- On Tue, 5/5/09, Rolf Turner <r.turner at auckland.ac.nz> wrote: