Variable importance via variable permutations

Usage

variable_importance(
  object,
  data = NULL,
  variables = NULL,
  response = NULL,
  loss_function = NULL,
  iterations = 1,
  sample_size = NULL,
  sample_frac = NULL,
  predict_function = NULL,
  parallel = FALSE,
  verbose = TRUE
)

Arguments

object: The model object.
data: A data to calculate the loss_function.
variables: Variables to use.
response: Name of the variable response.
loss_function: The loss function to evaluate, Must be a function with 2 arguments: actual and predicted values. Loss function gives a smaller value if the model have better performance of the model.
iterations: Number of iterations.
sample_size: Sample size.
sample_frac: Proportion to sample in each iteration.
predict_function: Predict function, usually is a function(model, newdata) which returns a vector (no data frame).
parallel: A logical value indicating if the process should be using furrr::future_pmap_dbl or purrr::pmap_dbl.
verbose: A logical value indicating to show progress bars.

Examples


lm_model <- lm(mpg ~ ., data = mtcars)

vi <- variable_importance(
  lm_model,
  response = "mpg",
  data = mtcars
)
#> ℹ Using all variables in data.
#> ℹ Using root mean square error as loss function.
#> ℹ Using `base::identity` as sampler.
#> ℹ Using `predict.lm` as predict function.

plot(vi)