roles: Manually alter roles

In recipes: Preprocessing and Feature Engineering Steps for Modeling

Manually alter roles

Description

update_role() alters an existing role in the recipe or assigns an initial role to variables that do not yet have a declared role.

add_role() adds an additional role to variables that already have a role in the recipe. It does not overwrite old roles, as a single variable can have multiple roles.

remove_role() eliminates a single existing role in the recipe.

Usage

add_role(recipe, ..., new_role = "predictor", new_type = NULL)

update_role(recipe, ..., new_role = "predictor", old_role = NULL)

remove_role(recipe, ..., old_role)

Arguments

recipe	An existing recipe().
...	One or more selector functions to choose which variables are being assigned a role. See selections() for more details.
new_role	A character string for a single role.
new_type	A character string for specific type that the variable should be identified as. If left as NULL, the type is automatically identified as the first type you see for that variable in summary(recipe).
old_role	A character string for the specific role to update for the variables selected by .... update_role() accepts a NULL as long as the variables have only a single role.

Details

update_role(), add_role() and remove_role() will be applied on a recipe before any of the steps or checks, regardless of where they are located in position. This means that roles can only be changed with these three functions for columns that are already present in the original data supplied to recipe(). See the role argument in some step functions to update roles for columns created by steps.

Variables can have any arbitrary role (see the examples) but there are three special standard roles, "predictor", "outcome", and "case_weights". The first two roles are typically required when fitting a model.

update_role() should be used when a variable doesn't currently have a role in the recipe, or to replace an old_role with a new_role. add_role() only adds additional roles to variables that already have roles and will throw an error when the current role is missing (i.e. NA).

When using add_role(), if a variable is selected that already has the new_role, a warning is emitted and that variable is skipped so no duplicate roles are added.

Adding or updating roles is a useful way to group certain variables that don't fall in the standard "predictor" bucket. You can perform a step on all of the variables that have a custom role with the selector has_role().

Effects of non-standard roles

Recipes can label and retain column(s) of your data set that should not be treated as outcomes or predictors. A unique identifier column or some other ancillary data could be used to troubleshoot issues during model development but may not be either an outcome or predictor.

For example, the modeldata::biomass dataset has a column named sample with information about the specific sample type. We can change that role:

library(recipes)

data(biomass, package = "modeldata")
biomass_train <- biomass[1:100,]
biomass_test <- biomass[101:200,]

rec <- recipe(HHV ~ ., data = biomass_train) %>%
  update_role(sample, new_role = "id variable") %>%
  step_center(carbon)

rec <- prep(rec, biomass_train)

This means that sample is no longer treated as a "predictor" (the default role for columns on the right-hand side of the formula supplied to recipe()) and won't be used in model fitting or analysis, but will still be retained in the data set.

If you really aren't using sample in your recipe, we recommend that you instead remove sample from your dataset before passing it to recipe(). The reason for this is because recipes assumes that all non-standard roles are required at bake() time (or predict() time, if you are using a workflow). Since you didn't use sample in any steps of the recipe, you might think that you don't need to pass it to bake(), but this isn't true because recipes doesn't know that you didn't use it:

biomass_test$sample <- NULL

bake(rec, biomass_test)
#> Error in `bake()`:
#> x The following required columns are missing from `new_data`: `sample`.
#> i These columns have one of the following roles, which are required at `bake()`
#>   time: `id variable`.
#> i If these roles are not required at `bake()` time, use
#>   `update_role_requirements(role = "your_role", bake = FALSE)`.

As we mentioned before, the best way to avoid this issue is to not even use a role, just remove the sample column from biomass before calling recipe(). In general, predictors and non-standard roles that are supplied to recipe() should be present at both prep() and bake() time.

If you can't remove sample for some reason, then the second best way to get around this issue is to tell recipes that the "id variable" role isn't required at bake() time. You can do that by using update_role_requirements():

rec <- recipe(HHV ~ ., data = biomass_train) %>%
  update_role(sample, new_role = "id variable") %>%
  update_role_requirements("id variable", bake = FALSE) %>%
  step_center(carbon)

rec <- prep(rec, biomass_train)

# No errors!
biomass_test_baked <- bake(rec, biomass_test)

It should be very rare that you need this feature.

Value

An updated recipe object.

Examples

library(recipes)
data(biomass, package = "modeldata")

# Using the formula method, roles are created for any outcomes and predictors:
recipe(HHV ~ ., data = biomass) %>%
  summary()

# However `sample` and `dataset` aren't predictors. Since they already have
# roles, `update_role()` can be used to make changes, to any arbitrary role:
recipe(HHV ~ ., data = biomass) %>%
  update_role(sample, new_role = "id variable") %>%
  update_role(dataset, new_role = "splitting variable") %>%
  summary()

# `update_role()` cannot set a role to NA, use `remove_role()` for that
## Not run: 
recipe(HHV ~ ., data = biomass) %>%
  update_role(sample, new_role = NA_character_)

## End(Not run)

# ------------------------------------------------------------------------------

# Variables can have more than one role. `add_role()` can be used
# if the column already has at least one role:
recipe(HHV ~ ., data = biomass) %>%
  add_role(carbon, sulfur, new_role = "something") %>%
  summary()

# `update_role()` has an argument called `old_role` that is required to
# unambiguously update a role when the column currently has multiple roles.
recipe(HHV ~ ., data = biomass) %>%
  add_role(carbon, new_role = "something") %>%
  update_role(carbon, new_role = "something else", old_role = "something") %>%
  summary()

# `carbon` has two roles at the end, so the last `update_roles()` fails since
# `old_role` was not given.
## Not run: 
recipe(HHV ~ ., data = biomass) %>%
  add_role(carbon, sulfur, new_role = "something") %>%
  update_role(carbon, new_role = "something else")

## End(Not run)

# ------------------------------------------------------------------------------

# To remove a role, `remove_role()` can be used to remove a single role.
recipe(HHV ~ ., data = biomass) %>%
  add_role(carbon, new_role = "something") %>%
  remove_role(carbon, old_role = "something") %>%
  summary()

# To remove all roles, call `remove_role()` multiple times to reset to `NA`
recipe(HHV ~ ., data = biomass) %>%
  add_role(carbon, new_role = "something") %>%
  remove_role(carbon, old_role = "something") %>%
  remove_role(carbon, old_role = "predictor") %>%
  summary()

# ------------------------------------------------------------------------------

# If the formula method is not used, all columns have a missing role:
recipe(biomass) %>%
  summary()

recipes documentation built on July 4, 2024, 9:06 a.m.