Fit an (relaxed) lasso or elastic net penalized regression via glmnet

A wrapper for glmnet and cv.glmnet. It runs CV by default so remember to set your seed for reproducibility.

Usage

select_glmnet(
  formula,
  data,
  family = c("gaussian", "binomial", "poisson"),
  lambda = c("best", "compact"),
  fitted_selector = NULL,
  ...
)

formula: a formula
data: data set
family: outcome distributional family
lambda: can be best, compact (which use CV), or a numeric vector. See details.
fitted_selector: a previously fit selector, used for resampling
...: Additional arguments that can be passed to glmnet, e.g. alpha

A selector object wrapping glmnet containing:

beta: a tibble containing term names and coefficients
std: Was desing matrix standadrized
penalty: penalty used (lasso or MCP)
lambda: are the coefficeint associated with "lambda.min" or "lambda.1se"
lambda.select: numeric value of selected lambda
fold: Which fold each observation belongs to. By default the observations are randomly assigned.
x: the model dataframe used
y: repsonse used
alpha: selected alpha for model fitting