Random Forest Variable Importance — rf_importance • mars

Returns aggregated split-based variable importance for a fitted mars_rf object.

Usage

rf_importance(object, scale = TRUE, ...)

Arguments

object: A fitted "mars_rf" object.
scale: Logical; if TRUE, return relative importance that sums to 1.
...: Not currently used.

Value

A data frame of variable importance values.

Examples

# \donttest{
rf_fit <- mars_rf_univariate(
  data = teacher_expectancy,
  formula = yi ~ year + weeks,
  studyID = "study",
  variance = "vi",
  num_trees = 25,
  seed = 123
)
rf_importance(rf_fit)
#>   predictor importance
#> 1     weeks 0.94495095
#> 2      year 0.05504905
# }