I am looking at the R function gausspr
from the kernlab
package for Gaussian process regression. The process is defined by the hyperparameters of the kernel function and by the noise in the data. I see in the documentation that I can specify
var: the initial noise variance, (only for regression) (default : 0.001)
but I do not see how to access the estimated value after the regression has run. For instance, consider I have some observed points, and want to predict y values at the locations given by X
:
obs <- data.frame(x = c(-4, -3, -1, 0, 2),
y = c(-2, 0, 1, 2, -1))
X <- seq(-5,5,len=50)
I can do so with kernlab::gausspr
as such:
gp <- gausspr(obs$x, obs$y, kernel="rbfdot", scaled=FALSE, var=.09)
Ef <- predict(gp, X)
I can get the estimated value of the kernel hyperparameter:
gp@kernelf@kpar
But I don't see how I can return the estimated value of the noise parameter, var
?