R help - May 2006 - Regression line limited by the rage of values

Hi

In R, using  plot(x,y)   followed by abline(lm(y~x)) produces a graph 
with a regression line spanning the whole plot . This means that the 
line extends beyond the swarm of data points to the defined of default  
plot region. With par(xpd=T) it will span the entire figure region. But 
how can I limit a regression line to the data range, i.e between  
(xmin,ymin) and  (xmax,ymax)?

Sorry for not knowing the lingo here. If you don't understand the 
question, please run the following script:

x1<-c(1,2,3,4)
x2<-c(5,6,7,8)
y1<-c(2,4,5,8)
y2<-c(10,11,12,16)
plot(x1,y1,xlim=c(0,10),ylim=c(0,20),col="blue")
points(x2,y2,col="red")
abline(lm(y1~x1),col="blue")
abline(lm(y2~x2),col="red")

The resulting plot isn't very informative. There is no overlap in the 
two groups of data, yet the two ablines overlap.
I instead  want the blue line to go from (1,2) to (4,8) and the red line 
from (5,10) to (8,16).

So, how can I constrain the abline to the relevant region, i.e stop 
abline from extrapolating beyond the actual range of data.
Or should I use a function line 'lines' to do this?

Cheers
Andres

On Wed, 2006-05-24 at 18:51 +0200, Andreas Svensson
wrote:
> Hi
> 
> In R, using  plot(x,y)   followed by abline(lm(y~x)) produces a graph 
> with a regression line spanning the whole plot . This means that the 
> line extends beyond the swarm of data points to the defined of default  
> plot region. With par(xpd=T) it will span the entire figure region. But 
> how can I limit a regression line to the data range, i.e between  
> (xmin,ymin) and  (xmax,ymax)?
> 
> Sorry for not knowing the lingo here. If you don't understand the 
> question, please run the following script:
> 
> x1<-c(1,2,3,4)
> x2<-c(5,6,7,8)
> y1<-c(2,4,5,8)
> y2<-c(10,11,12,16)
> plot(x1,y1,xlim=c(0,10),ylim=c(0,20),col="blue")
> points(x2,y2,col="red")
> abline(lm(y1~x1),col="blue")
> abline(lm(y2~x2),col="red")
> 
> The resulting plot isn't very informative. There is no overlap in the 
> two groups of data, yet the two ablines overlap.
> I instead  want the blue line to go from (1,2) to (4,8) and the red line 
> from (5,10) to (8,16).
> 
> So, how can I constrain the abline to the relevant region, i.e stop 
> abline from extrapolating beyond the actual range of data.
> Or should I use a function line 'lines' to do this?
> 
> Cheers
> Andres
Try this instead of the two abline()'s:

  lines(x1, fitted(lm(y1 ~ x1)), col = "blue")

  lines(x2, fitted(lm(y2 ~ x2)), col = "red")

The function fitted() will extract the model fitted y values from the
lm() model object.

See ?lines and ?fitted

HTH,

Marc Schwartz

For single lines the usual method is to just use the lines or segments
function, however I have been thinking that there may be some other uses
for clipping within a plot region when adding info (abline, but others
as well).  So here is a first stab at a function to clip within the
region, it needs the TeachingDemos package to work.  I'll probably put
in some more error checking, document it and include it in a future
TeachingDemos release, but you are welcome to use it in the meantime.

For your case you would create your scatterplot the same as you already
did, then add the lines like:
> clipplot( abline(lm(y1~x1),col="blue"), xlim=c(1,4) )
> clipplot( abline(lm(y2~x2),col="red"), xlim=c(5,8) )
Here is the function definition:

clipplot <- function(fun, xlim=par('usr')[1:2], 
	ylim=par('usr')[3:4] ){
  old.par <- par(c('plt','xpd'))


  if( length(xlim) < 2 ) stop('xlim must be a vector with at least 2
elements')
  if( length(ylim) < 2 ) stop('ylim must be a vector with at least 2
elements')

  if( !require(TeachingDemos) ) stop('TeachingDemos package needed')

  xl <- range(xlim)
  yl <- range(ylim)

  pc <- cnvrt.coords(xl,yl)$fig

  par(plt=c(pc$x,pc$y),xpd=FALSE)

  fun

  par(old.par)
  box() # need to plot something to reset

}  

 
Hope this helps,


-- 
Gregory (Greg) L. Snow Ph.D.
Statistical Data Center
Intermountain Healthcare
greg.snow at intermountainmail.org
(801) 408-8111
 

-----Original Message-----
From: r-help-bounces at stat.math.ethz.ch
[mailto:r-help-bounces at stat.math.ethz.ch] On Behalf Of Andreas Svensson
Sent: Wednesday, May 24, 2006 10:52 AM
To: r-help at stat.math.ethz.ch
Subject: [R] Regression line limited by the rage of values

Hi

In R, using  plot(x,y)   followed by abline(lm(y~x)) produces a graph 
with a regression line spanning the whole plot . This means that the
line extends beyond the swarm of data points to the defined of default
plot region. With par(xpd=T) it will span the entire figure region. But
how can I limit a regression line to the data range, i.e between
(xmin,ymin) and  (xmax,ymax)?

Sorry for not knowing the lingo here. If you don't understand the
question, please run the following script:

x1<-c(1,2,3,4)
x2<-c(5,6,7,8)
y1<-c(2,4,5,8)
y2<-c(10,11,12,16)
plot(x1,y1,xlim=c(0,10),ylim=c(0,20),col="blue")
points(x2,y2,col="red")
abline(lm(y1~x1),col="blue")
abline(lm(y2~x2),col="red")

The resulting plot isn't very informative. There is no overlap in the
two groups of data, yet the two ablines overlap.
I instead  want the blue line to go from (1,2) to (4,8) and the red line
from (5,10) to (8,16).

So, how can I constrain the abline to the relevant region, i.e stop
abline from extrapolating beyond the actual range of data.
Or should I use a function line 'lines' to do this?

Cheers
Andres

______________________________________________
R-help at stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide!
http://www.R-project.org/posting-guide.html

Andreas Svensson wrote:
> So, how can I constrain the abline to the relevant region, i.e stop
> abline from extrapolating beyond the actual range of data.
> Or should I use a function line 'lines' to do this?
One elegant way of doing this is using 'xyplot' from 'lattice'
and adding a
loess line with a large value of 'span' (which basically makes is equal
to
the ordinary least-squares regression line). Here's a simple example:

library(lattice)
data(iris)
xyplot(Petal.Length~Petal.Width,
       groups=Species, type=c("p","smooth"), span=100,
data=iris)

-- 
Karl Ove Hufthammer