CODEWITHSUNDEEP
rmgcv
generic_user
generic_user

Reputation: 3562

Save data from a gam plot without actually plotting the data?

Is there a convenient way to extract the data from a gam plot, without actually plotting the gam object?

Here's a dummy example. plot.data has the data I want in it, but I don't want the plot window to be affected.

library(mgcv)    
x=1:10000/1000
y = sin(x)+rnorm(10000,sd=2)
m = gam(y~s(x))
plot.data<-plot(m,plot=F)

Upvotes: 2

Views: 1006

Answers (2)

Rorschach
Rorschach

Reputation: 32426

It doesn't look like plot.gam has the option for not plotting. But you could try

plot.data <- {
    dev.new()
    res <- plot(m)
    dev.off()
    res
}

or possibly

plot.data <- {
    pdf(NULL)
    res <- plot(m)
    invisible(dev.off())
    res
}

Upvotes: 6

ulfelder
ulfelder

Reputation: 5335

If you're using gam() from the package gam, then you can get those data in list form by calling preplot(m). Here's what that looks like for your data:

library(gam)
x = 1:10000/1000
y = sin(x)+rnorm(10000,sd=2)
m = gam(y~s(x))
preplot(m)

List of 1
 $ s(x):List of 5
  ..$ x   : num [1:10000] 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 0.01 ...
  ..$ y   : Named num [1:10000] 0.421 0.421 0.421 0.421 0.421 ...
  .. ..- attr(*, "names")= chr [1:10000] "1" "2" "3" "4" ...
  ..$ se.y: Named num [1:10000] 0.0783 0.0782 0.0781 0.0781 0.078 ...
  .. ..- attr(*, "names")= chr [1:10000] "1" "2" "3" "4" ...
  ..$ xlab: chr "x"
  ..$ ylab: chr "s(x)"
  ..- attr(*, "class")= chr "preplot.gam"
 - attr(*, "class")= chr "preplot.gam"

The x and y components of that list are what I think you're after. Presumably, if you've got more than one smoothing term in your model, the ith component of the preplot list will correspond to the ith smoothing term in your original call to gam().

Upvotes: 1

Related Questions