workflows (version 1.2.0)

extract-workflow: Extract elements of a workflow

Description

These functions extract various elements from a workflow object. If they do not exist yet, an error is thrown.

• extract_preprocessor() returns the formula, recipe, or variable expressions used for preprocessing.

• extract_spec_parsnip() returns the parsnip model specification.

• extract_fit_parsnip() returns the parsnip model fit object.

• extract_fit_engine() returns the engine specific fit embedded within a parsnip model fit. For example, when using parsnip::linear_reg() with the "lm" engine, this returns the underlying lm object.

• extract_mold() returns the preprocessed "mold" object returned from hardhat::mold(). It contains information about the preprocessing, including either the prepped recipe, the formula terms object, or variable selectors.

• extract_recipe() returns the recipe. The estimated argument specifies whether the fitted or original recipe is returned.

• extract_parameter_dials() returns a single dials parameter object.

• extract_parameter_set_dials() returns a set of dials parameter objects.

• extract_fit_time() returns a tibble with elapsed fit times. The fit times correspond to the time for the parsnip engine or recipe steps to fit (or their sum if summarize = TRUE) and do not include other portions of the elapsed time in fit.workflow().

Usage

# S3 method for workflow
extract_spec_parsnip(x, ...)
# S3 method for workflow
extract_recipe(x, ..., estimated = TRUE)
# S3 method for workflow
extract_fit_parsnip(x, ...)
# S3 method for workflow
extract_fit_engine(x, ...)
# S3 method for workflow
extract_mold(x, ...)
# S3 method for workflow
extract_preprocessor(x, ...)
# S3 method for workflow
extract_parameter_set_dials(x, ...)
# S3 method for workflow
extract_parameter_dials(x, parameter, ...)
# S3 method for workflow
extract_fit_time(x, summarize = TRUE, ...)

Value

The extracted value from the object, x, as described in the description section.

Arguments

x

A workflow

...

Not currently used.

estimated

A logical for whether the original (unfit) recipe or the fitted recipe should be returned. This argument should be named.

parameter

A single string for the parameter ID.

summarize

A logical for whether the elapsed fit time should be returned as a single row or multiple rows.

Details

Extracting the underlying engine fit can be helpful for describing the model (via print(), summary(), plot(), etc.) or for variable importance/explainers.

However, users should not invoke the predict() method on an extracted model. There may be preprocessing operations that workflows has executed on the data prior to giving it to the model. Bypassing these can lead to errors or silently generating incorrect predictions.

Good:

workflow_fit %>% predict(new_data)

Bad:

workflow_fit %>% extract_fit_engine()  %>% predict(new_data)
# or
workflow_fit %>% extract_fit_parsnip() %>% predict(new_data)

Examples

library(parsnip)
library(recipes)
library(magrittr)

model <- linear_reg() %>%
  set_engine("lm")

recipe <- recipe(mpg ~ cyl + disp, mtcars) %>%
  step_log(disp)

base_wf <- workflow() %>%
  add_model(model)

recipe_wf <- add_recipe(base_wf, recipe)
formula_wf <- add_formula(base_wf, mpg ~ cyl + log(disp))
variable_wf <- add_variables(base_wf, mpg, c(cyl, disp))

fit_recipe_wf <- fit(recipe_wf, mtcars)
fit_formula_wf <- fit(formula_wf, mtcars)

# The preprocessor is a recipe, formula, or a list holding the
# tidyselect expressions identifying the outcomes/predictors
extract_preprocessor(recipe_wf)
extract_preprocessor(formula_wf)
extract_preprocessor(variable_wf)

# The `spec` is the parsnip spec before it has been fit.
# The `fit` is the fitted parsnip model.
extract_spec_parsnip(fit_formula_wf)
extract_fit_parsnip(fit_formula_wf)
extract_fit_engine(fit_formula_wf)

# The mold is returned from `hardhat::mold()`, and contains the
# predictors, outcomes, and information about the preprocessing
# for use on new data at `predict()` time.
extract_mold(fit_recipe_wf)

# A useful shortcut is to extract the fitted recipe from the workflow
extract_recipe(fit_recipe_wf)

# That is identical to
identical(
  extract_mold(fit_recipe_wf)$blueprint$recipe,
  extract_recipe(fit_recipe_wf)
)