2

I am using the glmnet package to get LASSO estimates, as follows:

model <- cv.glmnet(data, return, standardize = TRUE)

I am able to extract the coefficients using coef(model), however, I couldn't figure out a way of getting the standard error and Z-score for each variable.

halfer
  • 19,824
  • 17
  • 99
  • 186
Mayou
  • 8,498
  • 16
  • 59
  • 98
  • Looking at `str(model)` shows that `model$cvsd` gives the standard error of estimate of `cvm`. I don't see any other standard error here. – Metrics Sep 04 '13 at 14:44

1 Answers1

1

Since LASSO estimates are non-linear non-differentiable function of the response values, it is difficult to obtain an accurate estimate of their standard errors. The most common approach to derive the standard errors of LASSO estimates is to use bootstrapping the least squares fit standard errors.

Mayou
  • 8,498
  • 16
  • 59
  • 98