Discretize numeric variables with CART

step_discretize_cart creates a specification of a recipe step that will discretize numeric data (e.g. integers or doubles) into bins in a supervised way using a CART model.

step_discretize_cart(
  recipe,
  ...,
  role = NA,
  trained = FALSE,
  outcome = NULL,
  cost_complexity = 0.01,
  tree_depth = 10,
  min_n = 20,
  rules = NULL,
  skip = FALSE,
  id = rand_id("discretize_cart")
)

# S3 method for step_discretize_cart
tidy(x, ...)

Arguments

recipe	A recipe object. The step will be added to the sequence of operations for this recipe.
...	One or more selector functions to choose which variables are affected by the step. See selections() for more details.
role	Defaults to "predictor".
trained	A logical to indicate if the quantities for preprocessing have been estimated.
outcome	A call to vars to specify which variable is used as the outcome to train CART models in order to discretize explanatory variables.
cost_complexity	The regularization parameter. Any split that does not decrease the overall lack of fit by a factor of cost_complexity is not attempted. Corresponds to cp in rpart::rpart(). Defaults to 0.01.
tree_depth	The maximum depth in the final tree. Corresponds to maxdepth in rpart::rpart(). Defaults to 10.
min_n	The number of data points in a node required to continue splitting. Corresponds to minsplit in rpart::rpart(). Defaults to 20.
rules	The splitting rules of the best CART tree to retain for each variable. If length zero, splitting could not be used on that column.
skip	A logical. Should the step be skipped when the recipe is baked by recipes::bake.recipe()? While all operations are baked when recipes::prep.recipe() is run, some operations may not be able to be conducted on new data (e.g. processing the outcome variable(s)). Care should be taken when using skip = TRUE as it may affect the computations for subsequent operations
id	A character string that is unique to this step to identify it.
x	A step_discretize_cart object.

Value

An updated version of recipe with the new step added to the sequence of existing steps (if any).

Details

step_discretize_cart() creates non-uniform bins from numerical variables by utilizing the information about the outcome variable and applying a CART model.

The best selection of buckets for each variable is selected using the standard cost-complexity pruning of CART, which makes this discretization method resistant to overfitting.

This step requires the rpart package. If not installed, the step will stop with a note about installing the package.

Note that the original data will be replaced with the new bins.

See also

embed::step_discretize_xgb(), recipes::recipe(), recipes::prep.recipe(), recipes::bake.recipe()

Examples

library(modeldata)
data(ad_data)
library(rsample)

split <- initial_split(ad_data, strata = "Class")

ad_data_tr <- training(split)
ad_data_te <- testing(split)

cart_rec <-
  recipe(Class ~ ., data = ad_data_tr) %>%
  step_discretize_cart(tau, age, p_tau, Ab_42, outcome = "Class", id = "cart splits")

cart_rec <- prep(cart_rec, training = ad_data_tr)
#> Warning: `step_discretize_cart()` failed to find any meaningful splits for predictor 'age', which will not be binned.

# The splits: 
tidy(cart_rec, id = "cart splits")
#> # A tibble: 16 x 3
#>    terms values id         
#>    <chr>  <dbl> <chr>      
#>  1 tau     6.15 cart splits
#>  2 tau     6.25 cart splits
#>  3 tau     6.32 cart splits
#>  4 tau     6.42 cart splits
#>  5 tau     6.66 cart splits
#>  6 p_tau   3.90 cart splits
#>  7 p_tau   4.36 cart splits
#>  8 p_tau   4.40 cart splits
#>  9 p_tau   4.49 cart splits
#> 10 p_tau   4.54 cart splits
#> 11 p_tau   4.62 cart splits
#> 12 Ab_42   9.98 cart splits
#> 13 Ab_42  10.3  cart splits
#> 14 Ab_42  11.1  cart splits
#> 15 Ab_42  11.2  cart splits
#> 16 Ab_42  11.3  cart splits

bake(cart_rec, ad_data_te, tau)
#> # A tibble: 82 x 1
#>    tau          
#>    <fct>        
#>  1 [6.147,6.25) 
#>  2 [6.418,6.661)
#>  3 [-Inf,6.147) 
#>  4 [-Inf,6.147) 
#>  5 [6.147,6.25) 
#>  6 [-Inf,6.147) 
#>  7 [6.25,6.322) 
#>  8 [6.661, Inf] 
#>  9 [-Inf,6.147) 
#> 10 [-Inf,6.147) 
#> # … with 72 more rows