Safe predictions for cross-validated glmnet objects

# S3 method for glmnet
multi_predict(object, new_data, type = c("response",
  "class", "prob"), ..., params = NULL)

Arguments

object	TODO
new_data	TODO
type	TODO
...	Unused. `safe_predict()` checks that all arguments in `...` are evaluated via the `ellipsis` package. The idea is to prevent silent errors when arguments are mispelled. This feature is experimental and feedback is welcome.
params	works with a single value for the penalty. otherwise CV is gonna be inconsistent a wild pain in the rear. works with a vector of values for the penalty in the future will work with a tibble defining a parameter grid