[R] Annotate a segmented linear regression plot

[David Winsemius]

On Oct 2, 2012, at 3:38 PM, Ben Harrison wrote:

> On 28 September 2012 16:38, David Winsemius <dwinsemius at comcast.net> wrote:
>> 
>> ?text  # should be fairly clear.
> 
> Thank you. I was stupid to ask such a trivial question along with a
> not-so-trivial one. The second part of the question was probably more
> important: is there a way to obtain the location of segments produced
> by the segmented package, so that I can annotate them easily?

I guess it depends on what you mean by "the location of segments". It's easy enough to retrun the x-values for the knots:

    o$psi[ , "Est."]

I do not see a `predicted.segmented` function, so might need to construct separate lm.predict estimates for values between the pairs of x-values if you meant you wanted the y-values for the knots.

-- 
David.

> 
> I have since asked this question of Vito, the package's author. Here
> is his response:
> 
> #####
> dear Ben,
> here a possible solution,
> 
> o is the segmented fit
> 
>  r<-o$rangeZ[,1]
>  est.psi<-o$psi[,2]
>  v<-sort(c(r, est.psi))
>  xCoord<-rowMeans(cbind(v[-length(v)], v[-1]))
> 
>  Z<-o$model[,o$nameUV$Z]
> 
>  id<-sapply(xCoord, function(x)which.min(abs(x-Z)))
>  yCoord<-broken.line(o)[id]
> 
>  plot(o, col=2:4, res=T)
>  text(xCoord, yCoord, labels=formatC(slope(o)[[1]][,1], digits=2),
> pos=4, cex=1.3)
> 
> Play with pos, cex and digits to modify the appearance of the labels.
> 
> vito
> #####
> 
>> Please learn to post in plain text.
> 
> I hope this is better.
> 
> Ben.

David Winsemius, MD
Alameda, CA, USA