How to extract the CV errors for optimal lambda using glmnet package?

Question

I'm using the glment package for regression in R. I do the cross validation using cv.fit<-cv.glmnet(x,y,...), and I get optimum lambda using cvfit$lambda.min. but I want to also get the corresponduing MSE(mean square error) for that lambda. would someone help me to get it ?

Answer 1

From ?cv.glmnet:

# ...
# Value:
#
#     an object of class ‘"cv.glmnet"’ is returned, which is a list with
#     the ingredients of the cross-validation fit. 
#
# lambda: the values of ‘lambda’ used in the fits.
#
#   cvm: The mean cross-validated error - a vector of length
#       ‘length(lambda)’.
# ...

So in your case, the cross-validated mean squared errors are in cv.fit$cvm and the corresponding lambda values are in cv.fit$lambda.

To find the minimum MSE you can use which as follows:

i <- which(cv.fit$lambda == cv.fit$lambda.min)
mse.min <- cv.fit$cvm[i]

or shorter

mse.min <- cv.fit$cvm[cv.fit$lambda == cv.fit$lambda.min]

Answer 2

If you running glmnet with the loss function "mse", the minimum lambda represents the minimum MSE. Thus you could find it simply by:

mse.min <- min(cv.fit$cvm)