Package 'workflowsets'

Description

Use existing objects to create a workflow set. A list of objects that are either simple workflows or objects that have class "tune_results" can be converted into a workflow set.

Usage

as_workflow_set(...)

Arguments

Value

A workflow set. Note that the option column will not reflect the options that were used to create each object.

Note

The package supplies two pre-generated workflow sets, two_class_set and chi_features_set, and associated sets of model fits two_class_res and chi_features_res.

The ⁠two_class_*⁠ objects are based on a binary classification problem using the two_class_dat data from the modeldata package. The six models utilize either a bare formula or a basic recipe utilizing recipes::step_YeoJohnson() as a preprocessor, and a decision tree, logistic regression, or MARS model specification. See ?two_class_set for source code.

The ⁠chi_features_*⁠ objects are based on a regression problem using the Chicago data from the modeldata package. Each of the three models utilize a linear regression model specification, with three different recipes of varying complexity. The objects are meant to approximate the sequence of models built in Section 1.3 of Kuhn and Johnson (2019). See ?chi_features_set for source code.

Examples

# ------------------------------------------------------------------------------
# Existing results

# Use the already worked example to show how to add tuned
# objects to a workflow set
two_class_res

results <- two_class_res %>% purrr::pluck("result")
names(results) <- two_class_res$wflow_id

# These are all objects that have been resampled or tuned:
purrr::map_chr(results, ~ class(.x)[1])

# Use rlang's !!! operator to splice in the elements of the list
new_set <- as_workflow_set(!!!results)

# ------------------------------------------------------------------------------
# Make a set from unfit workflows

library(parsnip)
library(workflows)

lr_spec <- logistic_reg()

main_effects <-
  workflow() %>%
  add_model(lr_spec) %>%
  add_formula(Class ~ .)

interactions <-
  workflow() %>%
  add_model(lr_spec) %>%
  add_formula(Class ~ (.)^2)

as_workflow_set(main = main_effects, int = interactions)

Description

This autoplot() method plots performance metrics that have been ranked using a metric. It can also run autoplot() on the individual results (per wflow_id).

Usage

## S3 method for class 'workflow_set'
autoplot(
  object,
  rank_metric = NULL,
  metric = NULL,
  id = "workflow_set",
  select_best = FALSE,
  std_errs = qnorm(0.95),
  type = "class",
  ...
)

Arguments

Details

This function is intended to produce a default plot to visualize helpful information across all possible applications of a workflow set. A more appropriate plot for your specific analysis can be created by calling rank_results() and using standard ggplot2 code for plotting.

The x-axis is the workflow rank in the set (a value of one being the best) versus the performance metric(s) on the y-axis. With multiple metrics, there will be facets for each metric.

If multiple resamples are used, confidence bounds are shown for each result (90% confidence, by default).

Value

A ggplot object.

Note

The package supplies two pre-generated workflow sets, two_class_set and chi_features_set, and associated sets of model fits two_class_res and chi_features_res.

The ⁠two_class_*⁠ objects are based on a binary classification problem using the two_class_dat data from the modeldata package. The six models utilize either a bare formula or a basic recipe utilizing recipes::step_YeoJohnson() as a preprocessor, and a decision tree, logistic regression, or MARS model specification. See ?two_class_set for source code.

The ⁠chi_features_*⁠ objects are based on a regression problem using the Chicago data from the modeldata package. Each of the three models utilize a linear regression model specification, with three different recipes of varying complexity. The objects are meant to approximate the sequence of models built in Section 1.3 of Kuhn and Johnson (2019). See ?chi_features_set for source code.

Examples

autoplot(two_class_res)
autoplot(two_class_res, select_best = TRUE)
autoplot(two_class_res, id = "yj_trans_cart", metric = "roc_auc")

Description

The package supplies two pre-generated workflow sets, two_class_set and chi_features_set, and associated sets of model fits two_class_res and chi_features_res.

The ⁠two_class_*⁠ objects are based on a binary classification problem using the two_class_dat data from the modeldata package. The six models utilize either a bare formula or a basic recipe utilizing recipes::step_YeoJohnson() as a preprocessor, and a decision tree, logistic regression, or MARS model specification. See ?two_class_set for source code.

The ⁠chi_features_*⁠ objects are based on a regression problem using the Chicago data from the modeldata package. Each of the three models utilize a linear regression model specification, with three different recipes of varying complexity. The objects are meant to approximate the sequence of models built in Section 1.3 of Kuhn and Johnson (2019). See ?chi_features_set for source code.

Details

See below for the source code to generate the Chicago Features example workflow sets:

library(workflowsets)
library(workflows)
library(modeldata)
library(recipes)
library(parsnip)
library(dplyr)
library(rsample)
library(tune)
library(yardstick)
library(dials)

# ------------------------------------------------------------------------------
# Slightly smaller data size
data(Chicago)
Chicago <- Chicago[1:1195,]

time_val_split <-
   sliding_period(
      Chicago,
      date,
      "month",
      lookback = 38,
      assess_stop = 1
   )

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

base_recipe <-
   recipe(ridership ~ ., data = Chicago) %>%
   # create date features
   step_date(date) %>%
   step_holiday(date) %>%
   # remove date from the list of predictors
   update_role(date, new_role = "id") %>%
   # create dummy variables from factor columns
   step_dummy(all_nominal()) %>%
   # remove any columns with a single unique value
   step_zv(all_predictors()) %>%
   step_normalize(all_predictors())

date_only <-
   recipe(ridership ~ ., data = Chicago) %>%
   # create date features
   step_date(date) %>%
   update_role(date, new_role = "id") %>%
   # create dummy variables from factor columns
   step_dummy(all_nominal()) %>%
   # remove any columns with a single unique value
   step_zv(all_predictors())

date_and_holidays <-
   recipe(ridership ~ ., data = Chicago) %>%
   # create date features
   step_date(date) %>%
   step_holiday(date) %>%
   # remove date from the list of predictors
   update_role(date, new_role = "id") %>%
   # create dummy variables from factor columns
   step_dummy(all_nominal()) %>%
   # remove any columns with a single unique value
   step_zv(all_predictors())

date_and_holidays_and_pca <-
   recipe(ridership ~ ., data = Chicago) %>%
   # create date features
   step_date(date) %>%
   step_holiday(date) %>%
   # remove date from the list of predictors
   update_role(date, new_role = "id") %>%
   # create dummy variables from factor columns
   step_dummy(all_nominal()) %>%
   # remove any columns with a single unique value
   step_zv(all_predictors()) %>%
   step_pca(!!stations, num_comp = tune())

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

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

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

pca_param <-
   parameters(num_comp()) %>%
   update(num_comp = num_comp(c(0, 20)))

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

chi_features_set <-
   workflow_set(
      preproc = list(date = date_only,
                     plus_holidays = date_and_holidays,
                     plus_pca = date_and_holidays_and_pca),
      models = list(lm = lm_spec),
      cross = TRUE
   )

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

chi_features_res <-
   chi_features_set %>%
   option_add(param_info = pca_param, id = "plus_pca_lm") %>%
   workflow_map(resamples = time_val_split, grid = 21, seed = 1, verbose = TRUE)

References

Max Kuhn and Kjell Johnson (2019) Feature Engineering and Selection, https://bookdown.org/max/FES/a-more-complex-example.html

Examples

data(chi_features_set)

chi_features_set

Description

Return a tibble of performance metrics for all models or submodels.

Usage

## S3 method for class 'workflow_set'
collect_metrics(x, ..., summarize = TRUE)

## S3 method for class 'workflow_set'
collect_predictions(
  x,
  ...,
  summarize = TRUE,
  parameters = NULL,
  select_best = FALSE,
  metric = NULL
)

## S3 method for class 'workflow_set'
collect_notes(x, ...)

## S3 method for class 'workflow_set'
collect_extracts(x, ...)

Arguments

Details

When applied to a workflow set, the metrics and predictions that are returned do not contain the actual tuning parameter columns and values (unlike when these collect functions are run on other objects). The reason is that workflow sets can contain different types of models or models with different tuning parameters.

If the columns are needed, there are two options. First, the .config column can be used to merge the tuning parameter columns into an appropriate object. Alternatively, the map() function can be used to get the metrics from the original objects (see the example below).

Value

A tibble.

Note

The package supplies two pre-generated workflow sets, two_class_set and chi_features_set, and associated sets of model fits two_class_res and chi_features_res.

The ⁠two_class_*⁠ objects are based on a binary classification problem using the two_class_dat data from the modeldata package. The six models utilize either a bare formula or a basic recipe utilizing recipes::step_YeoJohnson() as a preprocessor, and a decision tree, logistic regression, or MARS model specification. See ?two_class_set for source code.

The ⁠chi_features_*⁠ objects are based on a regression problem using the Chicago data from the modeldata package. Each of the three models utilize a linear regression model specification, with three different recipes of varying complexity. The objects are meant to approximate the sequence of models built in Section 1.3 of Kuhn and Johnson (2019). See ?chi_features_set for source code.

See Also

tune::collect_metrics(), rank_results()

Examples

library(dplyr)
library(purrr)
library(tidyr)

two_class_res

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

collect_metrics(two_class_res)

# Alternatively, if the tuning parameter values are needed:
two_class_res %>%
  dplyr::filter(grepl("cart", wflow_id)) %>%
  mutate(metrics = map(result, collect_metrics)) %>%
  dplyr::select(wflow_id, metrics) %>%
  tidyr::unnest(cols = metrics)


collect_metrics(two_class_res, summarize = FALSE)

Description

comment_add() can be used to log important information about the workflow or its results as you work. Comments can be appended or removed.

Usage

comment_add(x, id, ..., append = TRUE, collapse = "\n")

comment_get(x, id)

comment_reset(x, id)

comment_print(x, id = NULL, ...)

Arguments

Value

comment_add() and comment_reset() return an updated workflow set. comment_get() returns a character string. comment_print() returns NULL invisibly.

Examples

two_class_set

two_class_set %>% comment_get("none_cart")

new_set <-
  two_class_set %>%
  comment_add("none_cart", "What does 'cart' stand for\u2753") %>%
  comment_add("none_cart", "Classification And Regression Trees.")

comment_print(new_set)

new_set %>% comment_get("none_cart")

new_set %>%
  comment_reset("none_cart") %>%
  comment_get("none_cart")

Description

These functions extract various elements from a workflow set 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_workflow_set_result() returns the results of workflow_map() for a particular workflow.

• extract_workflow() returns the workflow object. The workflow will not have been estimated.

• extract_parameter_set_dials() returns the parameter set that will be used to fit the supplied row id of the workflow set. Note that workflow sets reference a parameter set associated with the workflow contained in the info column by default, but can be fitted with a modified parameter set via the option_add() interface. This extractor returns the latter, if it exists, and returns the former if not, mirroring the process that workflow_map() follows to provide tuning functions a parameter set.

• extract_parameter_dials() returns the parameters object that will be used to fit the supplied tuning parameter in the supplied row id of the workflow set. See the above notes in extract_parameter_set_dials() on precedence.

Usage

extract_workflow_set_result(x, id, ...)

## S3 method for class 'workflow_set'
extract_workflow(x, id, ...)

## S3 method for class 'workflow_set'
extract_spec_parsnip(x, id, ...)

## S3 method for class 'workflow_set'
extract_recipe(x, id, ..., estimated = TRUE)

## S3 method for class 'workflow_set'
extract_fit_parsnip(x, id, ...)

## S3 method for class 'workflow_set'
extract_fit_engine(x, id, ...)

## S3 method for class 'workflow_set'
extract_mold(x, id, ...)

## S3 method for class 'workflow_set'
extract_preprocessor(x, id, ...)

## S3 method for class 'workflow_set'
extract_parameter_set_dials(x, id, ...)

## S3 method for class 'workflow_set'
extract_parameter_dials(x, id, parameter, ...)

Arguments

Details

These functions supersede the ⁠pull_*()⁠ functions (e.g., extract_workflow_set_result()).

Value

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

Note

The package supplies two pre-generated workflow sets, two_class_set and chi_features_set, and associated sets of model fits two_class_res and chi_features_res.

The ⁠two_class_*⁠ objects are based on a binary classification problem using the two_class_dat data from the modeldata package. The six models utilize either a bare formula or a basic recipe utilizing recipes::step_YeoJohnson() as a preprocessor, and a decision tree, logistic regression, or MARS model specification. See ?two_class_set for source code.

The ⁠chi_features_*⁠ objects are based on a regression problem using the Chicago data from the modeldata package. Each of the three models utilize a linear regression model specification, with three different recipes of varying complexity. The objects are meant to approximate the sequence of models built in Section 1.3 of Kuhn and Johnson (2019). See ?chi_features_set for source code.

Examples

library(tune)

two_class_res

extract_workflow_set_result(two_class_res, "none_cart")

extract_workflow(two_class_res, "none_cart")

Description

fit_best() takes results from tuning many models and fits the workflow configuration associated with the best performance to the training set.

Usage

## S3 method for class 'workflow_set'
fit_best(x, metric = NULL, eval_time = NULL, ...)

Arguments

Details

This function is a shortcut for the steps needed to fit the numerically optimal configuration in a fitted workflow set. The function ranks results, extracts the tuning result pertaining to the best result, and then again calls fit_best() (itself a wrapper) on the tuning result containing the best result.

In pseudocode:

rankings <- rank_results(wf_set, metric, select_best = TRUE)
tune_res <- extract_workflow_set_result(wf_set, rankings$wflow_id[1])
fit_best(tune_res, metric)

Note

The package supplies two pre-generated workflow sets, two_class_set and chi_features_set, and associated sets of model fits two_class_res and chi_features_res.

The ⁠two_class_*⁠ objects are based on a binary classification problem using the two_class_dat data from the modeldata package. The six models utilize either a bare formula or a basic recipe utilizing recipes::step_YeoJohnson() as a preprocessor, and a decision tree, logistic regression, or MARS model specification. See ?two_class_set for source code.

The ⁠chi_features_*⁠ objects are based on a regression problem using the Chicago data from the modeldata package. Each of the three models utilize a linear regression model specification, with three different recipes of varying complexity. The objects are meant to approximate the sequence of models built in Section 1.3 of Kuhn and Johnson (2019). See ?chi_features_set for source code.

Examples

library(tune)
library(modeldata)
library(rsample)

data(Chicago)
Chicago <- Chicago[1:1195, ]

time_val_split <-
  sliding_period(
    Chicago,
    date,
    "month",
    lookback = 38,
    assess_stop = 1
  )

chi_features_set

chi_features_res_new <-
  chi_features_set %>%
  # note: must set `save_workflow = TRUE` to use `fit_best()`
  option_add(control = control_grid(save_workflow = TRUE)) %>%
  # evaluate with resamples
  workflow_map(resamples = time_val_split, grid = 21, seed = 1, verbose = TRUE)

chi_features_res_new

# sort models by performance metrics
rank_results(chi_features_res_new)

# fit the numerically optimal configuration to the training set
chi_features_wf <- fit_best(chi_features_res_new)

chi_features_wf

# to select optimal value based on a specific metric:
fit_best(chi_features_res_new, metric = "rmse")

Description

From an initial model formula, create a list of formulas that exclude each predictor.

Usage

leave_var_out_formulas(formula, data, full_model = TRUE, ...)

Arguments

Details

The new formulas obey the hierarchy rule so that interactions without main effects are not included (unless the original formula contains such terms).

Factor predictors are left as-is (i.e., no indicator variables are created).

Value

A named list of formulas

See Also

workflow_set()

Examples

data(penguins, package = "modeldata")

leave_var_out_formulas(
  bill_length_mm ~ .,
  data = penguins
)

leave_var_out_formulas(
  bill_length_mm ~ (island + sex)^2 + flipper_length_mm,
  data = penguins
)

leave_var_out_formulas(
  bill_length_mm ~ (island + sex)^2 + flipper_length_mm +
    I(flipper_length_mm^2),
  data = penguins
)

Description

The option column controls options for the functions that are used to evaluate the workflow set, such as tune::fit_resamples() or tune::tune_grid(). Examples of common options to set for these functions include param_info and grid.

These functions are helpful for manipulating the information in the option column.

Usage

option_add(x, ..., id = NULL, strict = FALSE)

option_remove(x, ...)

option_add_parameters(x, id = NULL, strict = FALSE)

Arguments

Details

option_add() is used to update all of the options in a workflow set.

option_remove() will eliminate specific options across rows.

option_add_parameters() adds a parameter object to the option column (if parameters are being tuned).

Note that executing a function on the workflow set, such as tune_grid(), will add any options given to that function to the option column.

These functions do not control options for the individual workflows, such as the recipe blueprint. When creating a workflow manually, use workflows::add_model() or workflows::add_recipe() to specify extra options. To alter these in a workflow set, use update_workflow_model() or update_workflow_recipe().

Value

An updated workflow set.

Examples

library(tune)

two_class_set

two_class_set %>%
  option_add(grid = 10)

two_class_set %>%
  option_add(grid = 10) %>%
  option_add(grid = 50, id = "none_cart")

two_class_set %>%
  option_add_parameters()

Description

This function returns a named list with an extra class of "workflow_set_options" that has corresponding formatting methods for printing inside of tibbles.

Usage

option_list(...)

Arguments

Value

A classed list.

Examples

option_list(a = 1, b = 2)
option_list()

Description

Usage

pull_workflow_set_result(x, id)

pull_workflow(x, id)

Arguments

Details

pull_workflow_set_result() retrieves the results of workflow_map() for a particular workflow while pull_workflow() extracts the unfitted workflow from the info column.

The extract_workflow_set_result() and extract_workflow() functions should be used instead of these functions.

Value

pull_workflow_set_result() produces a tune_result or resample_results object. pull_workflow() returns an unfit workflow object.

Examples

library(tune)

two_class_res

pull_workflow_set_result(two_class_res, "none_cart")

pull_workflow(two_class_res, "none_cart")

Description

This function sorts the results by a specific performance metric.

Usage

rank_results(x, rank_metric = NULL, eval_time = NULL, select_best = FALSE)

Arguments

Details

If some models have the exact same performance, rank(value, ties.method = "random") is used (with a reproducible seed) so that all ranks are integers.

No columns are returned for the tuning parameters since they are likely to be different (or not exist) for some models. The wflow_id and .config columns can be used to determine the corresponding parameter values.

Value

A tibble with columns: wflow_id, .config, .metric, mean, std_err, n, preprocessor, model, and rank.

Note

The package supplies two pre-generated workflow sets, two_class_set and chi_features_set, and associated sets of model fits two_class_res and chi_features_res.

The ⁠two_class_*⁠ objects are based on a binary classification problem using the two_class_dat data from the modeldata package. The six models utilize either a bare formula or a basic recipe utilizing recipes::step_YeoJohnson() as a preprocessor, and a decision tree, logistic regression, or MARS model specification. See ?two_class_set for source code.

The ⁠chi_features_*⁠ objects are based on a regression problem using the Chicago data from the modeldata package. Each of the three models utilize a linear regression model specification, with three different recipes of varying complexity. The objects are meant to approximate the sequence of models built in Section 1.3 of Kuhn and Johnson (2019). See ?chi_features_set for source code.

Examples

chi_features_res

rank_results(chi_features_res)
rank_results(chi_features_res, select_best = TRUE)
rank_results(chi_features_res, rank_metric = "rsq")

Description

The package supplies two pre-generated workflow sets, two_class_set and chi_features_set, and associated sets of model fits two_class_res and chi_features_res.

The ⁠two_class_*⁠ objects are based on a binary classification problem using the two_class_dat data from the modeldata package. The six models utilize either a bare formula or a basic recipe utilizing recipes::step_YeoJohnson() as a preprocessor, and a decision tree, logistic regression, or MARS model specification. See ?two_class_set for source code.

The ⁠chi_features_*⁠ objects are based on a regression problem using the Chicago data from the modeldata package. Each of the three models utilize a linear regression model specification, with three different recipes of varying complexity. The objects are meant to approximate the sequence of models built in Section 1.3 of Kuhn and Johnson (2019). See ?chi_features_set for source code.

Details

See below for the source code to generate the Two Class example workflow sets:

library(workflowsets)
library(workflows)
library(modeldata)
library(recipes)
library(parsnip)
library(dplyr)
library(rsample)
library(tune)
library(yardstick)

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

data(two_class_dat, package = "modeldata")

set.seed(1)
folds <- vfold_cv(two_class_dat, v = 5)

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

decision_tree_rpart_spec <-
  decision_tree(min_n = tune(), cost_complexity = tune()) %>%
  set_engine('rpart') %>%
  set_mode('classification')

logistic_reg_glm_spec <-
  logistic_reg() %>%
  set_engine('glm')

mars_earth_spec <-
  mars(prod_degree = tune()) %>%
  set_engine('earth') %>%
  set_mode('classification')

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

yj_recipe <-
   recipe(Class ~ ., data = two_class_dat) %>%
   step_YeoJohnson(A, B)

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

two_class_set <-
   workflow_set(
      preproc = list(none = Class ~ A + B, yj_trans = yj_recipe),
      models = list(cart = decision_tree_rpart_spec, glm = logistic_reg_glm_spec,
                    mars = mars_earth_spec)
   )

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

two_class_res <-
  two_class_set %>%
  workflow_map(
    resamples = folds,
    grid = 10,
    seed = 2,
    verbose = TRUE,
    control = control_grid(save_workflow = TRUE)
  )

Examples

data(two_class_set)

two_class_set

Description

Workflows can take special arguments for the recipe (e.g. a blueprint) or a model (e.g. a special formula). However, when creating a workflow set, there is no way to specify these extra components.

update_workflow_model() and update_workflow_recipe() allow users to set these values after the workflow set is initially created. They are analogous to workflows::add_model() or workflows::add_recipe().

Usage

update_workflow_model(x, id, spec, formula = NULL)

update_workflow_recipe(x, id, recipe, blueprint = NULL)

Arguments

Note

The package supplies two pre-generated workflow sets, two_class_set and chi_features_set, and associated sets of model fits two_class_res and chi_features_res.

The ⁠two_class_*⁠ objects are based on a binary classification problem using the two_class_dat data from the modeldata package. The six models utilize either a bare formula or a basic recipe utilizing recipes::step_YeoJohnson() as a preprocessor, and a decision tree, logistic regression, or MARS model specification. See ?two_class_set for source code.

The ⁠chi_features_*⁠ objects are based on a regression problem using the Chicago data from the modeldata package. Each of the three models utilize a linear regression model specification, with three different recipes of varying complexity. The objects are meant to approximate the sequence of models built in Section 1.3 of Kuhn and Johnson (2019). See ?chi_features_set for source code.

Examples

library(parsnip)

new_mod <-
  decision_tree() %>%
  set_engine("rpart", method = "anova") %>%
  set_mode("classification")

new_set <- update_workflow_model(two_class_res, "none_cart", spec = new_mod)

new_set

extract_workflow(new_set, id = "none_cart")

Description

workflow_map() will execute the same function across the workflows in the set. The various ⁠tune_*()⁠ functions can be used as well as tune::fit_resamples().

Usage

workflow_map(
  object,
  fn = "tune_grid",
  verbose = FALSE,
  seed = sample.int(10^4, 1),
  ...
)

Arguments

Details

When passing options, anything passed in the ... will be combined with any values in the option column. The values in ... will override that column's values and the new options are added to the options column.

Any failures in execution result in the corresponding row of results to contain a try-error object.

In cases where a model has no tuning parameters is mapped to one of the tuning functions, tune::fit_resamples() will be used instead and a warning is issued if verbose = TRUE.

If a workflow requires packages that are not installed, a message is printed and workflow_map() continues with the next workflow (if any).

Value

An updated workflow set. The option column will be updated with any options for the tune package functions given to workflow_map(). Also, the results will be added to the result column. If the computations for a workflow fail, a try-catch object will be saved in place of the results (without stopping execution).

Note

The package supplies two pre-generated workflow sets, two_class_set and chi_features_set, and associated sets of model fits two_class_res and chi_features_res.

The ⁠two_class_*⁠ objects are based on a binary classification problem using the two_class_dat data from the modeldata package. The six models utilize either a bare formula or a basic recipe utilizing recipes::step_YeoJohnson() as a preprocessor, and a decision tree, logistic regression, or MARS model specification. See ?two_class_set for source code.

The ⁠chi_features_*⁠ objects are based on a regression problem using the Chicago data from the modeldata package. Each of the three models utilize a linear regression model specification, with three different recipes of varying complexity. The objects are meant to approximate the sequence of models built in Section 1.3 of Kuhn and Johnson (2019). See ?chi_features_set for source code.

See Also

workflow_set(), as_workflow_set(), extract_workflow_set_result()

Examples

library(workflowsets)
library(workflows)
library(modeldata)
library(recipes)
library(parsnip)
library(dplyr)
library(rsample)
library(tune)
library(yardstick)
library(dials)

# An example of processed results
chi_features_res

# Recreating them:

# ---------------------------------------------------------------------------
data(Chicago)
Chicago <- Chicago[1:1195, ]

time_val_split <-
  sliding_period(
    Chicago,
    date,
    "month",
    lookback = 38,
    assess_stop = 1
  )

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

base_recipe <-
  recipe(ridership ~ ., data = Chicago) %>%
  # create date features
  step_date(date) %>%
  step_holiday(date) %>%
  # remove date from the list of predictors
  update_role(date, new_role = "id") %>%
  # create dummy variables from factor columns
  step_dummy(all_nominal()) %>%
  # remove any columns with a single unique value
  step_zv(all_predictors()) %>%
  step_normalize(all_predictors())

date_only <-
  recipe(ridership ~ ., data = Chicago) %>%
  # create date features
  step_date(date) %>%
  update_role(date, new_role = "id") %>%
  # create dummy variables from factor columns
  step_dummy(all_nominal()) %>%
  # remove any columns with a single unique value
  step_zv(all_predictors())

date_and_holidays <-
  recipe(ridership ~ ., data = Chicago) %>%
  # create date features
  step_date(date) %>%
  step_holiday(date) %>%
  # remove date from the list of predictors
  update_role(date, new_role = "id") %>%
  # create dummy variables from factor columns
  step_dummy(all_nominal()) %>%
  # remove any columns with a single unique value
  step_zv(all_predictors())

date_and_holidays_and_pca <-
  recipe(ridership ~ ., data = Chicago) %>%
  # create date features
  step_date(date) %>%
  step_holiday(date) %>%
  # remove date from the list of predictors
  update_role(date, new_role = "id") %>%
  # create dummy variables from factor columns
  step_dummy(all_nominal()) %>%
  # remove any columns with a single unique value
  step_zv(all_predictors()) %>%
  step_pca(!!stations, num_comp = tune())

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

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

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

pca_param <-
  parameters(num_comp()) %>%
  update(num_comp = num_comp(c(0, 20)))

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

chi_features_set <-
  workflow_set(
    preproc = list(
      date = date_only,
      plus_holidays = date_and_holidays,
      plus_pca = date_and_holidays_and_pca
    ),
    models = list(lm = lm_spec),
    cross = TRUE
  )

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

chi_features_res_new <-
  chi_features_set %>%
  option_add(param_info = pca_param, id = "plus_pca_lm") %>%
  workflow_map(resamples = time_val_split, grid = 21, seed = 1, verbose = TRUE)

chi_features_res_new

Description

Often a data practitioner needs to consider a large number of possible modeling approaches for a task at hand, especially for new data sets and/or when there is little knowledge about what modeling strategy will work best. Workflow sets provide an expressive interface for investigating multiple models or feature engineering strategies in such a situation.

Usage

workflow_set(preproc, models, cross = TRUE, case_weights = NULL)

Arguments

Details

The preprocessors that can be combined with the model objects can be one or more of:

• A traditional R formula.

• A recipe definition (un-prepared) via recipes::recipe().

• A selectors object created by workflows::workflow_variables().

Since preproc is a named list column, any combination of these can be used in that argument (i.e., preproc can be mixed types).

Value

A tibble with extra class 'workflow_set'. A new set includes four columns (but others can be added):

• wflow_id contains character strings for the preprocessor/workflow combination. These can be changed but must be unique.

• info is a list column with tibbles containing more specific information, including any comments added using comment_add(). This tibble also contains the workflow object (which can be easily retrieved using extract_workflow()).

• option is a list column that will include a list of optional arguments passed to the functions from the tune package. They can be added manually via option_add() or automatically when options are passed to workflow_map().

• result is a list column that will contain any objects produced when workflow_map() is used.

Case weights

The case_weights argument can be passed as a single unquoted column name identifying the data column giving model case weights. For each workflow in the workflow set using an engine that supports case weights, the case weights will be added with workflows::add_case_weights(). workflow_set() will warn if any of the workflows specify an engine that does not support case weights—and ignore the case weights argument for those workflows—but will not fail.

Read more about case weights in the tidymodels at ?parsnip::case_weights.

Note

The package supplies two pre-generated workflow sets, two_class_set and chi_features_set, and associated sets of model fits two_class_res and chi_features_res.

The ⁠two_class_*⁠ objects are based on a binary classification problem using the two_class_dat data from the modeldata package. The six models utilize either a bare formula or a basic recipe utilizing recipes::step_YeoJohnson() as a preprocessor, and a decision tree, logistic regression, or MARS model specification. See ?two_class_set for source code.

The ⁠chi_features_*⁠ objects are based on a regression problem using the Chicago data from the modeldata package. Each of the three models utilize a linear regression model specification, with three different recipes of varying complexity. The objects are meant to approximate the sequence of models built in Section 1.3 of Kuhn and Johnson (2019). See ?chi_features_set for source code.

See Also

workflow_map(), comment_add(), option_add(), as_workflow_set()

Examples

library(workflowsets)
library(workflows)
library(modeldata)
library(recipes)
library(parsnip)
library(dplyr)
library(rsample)
library(tune)
library(yardstick)

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

data(cells)
cells <- cells %>% dplyr::select(-case)

set.seed(1)
val_set <- validation_split(cells)

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

basic_recipe <-
  recipe(class ~ ., data = cells) %>%
  step_YeoJohnson(all_predictors()) %>%
  step_normalize(all_predictors())

pca_recipe <-
  basic_recipe %>%
  step_pca(all_predictors(), num_comp = tune())

ss_recipe <-
  basic_recipe %>%
  step_spatialsign(all_predictors())

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

knn_mod <-
  nearest_neighbor(neighbors = tune(), weight_func = tune()) %>%
  set_engine("kknn") %>%
  set_mode("classification")

lr_mod <-
  logistic_reg() %>%
  set_engine("glm")

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

preproc <- list(none = basic_recipe, pca = pca_recipe, sp_sign = ss_recipe)
models <- list(knn = knn_mod, logistic = lr_mod)

cell_set <- workflow_set(preproc, models, cross = TRUE)
cell_set

# ------------------------------------------------------------------------------
# Using variables and formulas

# Select predictors by their names
channels <- paste0("ch_", 1:4)
preproc <- purrr::map(channels, ~ workflow_variables(class, c(contains(!!.x))))
names(preproc) <- channels
preproc$everything <- class ~ .
preproc

cell_set_by_group <- workflow_set(preproc, models["logistic"])
cell_set_by_group