Package 'bundle' reference manual

Title:	Serialize Model Objects with a Consistent Interface
Description:	Typically, models in 'R' exist in memory and can be saved via regular 'R' serialization. However, some models store information in locations that cannot be saved using 'R' serialization alone. The goal of 'bundle' is to provide a common interface to capture this information, situate it within a portable object, and restore it for use in new settings.
Authors:	Julia Silge [aut, cre] , Simon Couch [aut], Qiushi Yan [aut], Max Kuhn [aut], Posit Software, PBC [cph, fnd]
Maintainer:	Julia Silge <[email protected]>
License:	MIT + file LICENSE
Version:	0.1.2.9000
Built:	2024-12-12 06:30:38 UTC
Source:	https://github.com/rstudio/bundle

Bundling

Description

bundle() methods provide a consistent interface to serialization methods for statistical model objects. The created bundle can be saved, then re-loaded and unbundle()d in a new R session for use in prediction.

Usage

bundle(x, ...)

unbundle(x)
bundle(x, ...)

unbundle(x)

Arguments

`x`	A model object to bundle.
`...`	Additional arguments to bundle methods.

Value

A bundle object with subclass referencing the modeling function. If a bundle method is not defined for the supplied object, bundle.default is the identity function.

Bundles are a list subclass with two components:

object

An R object. Gives the output of native serialization methods from the model-supplying package, sometimes with additional classes or attributes that aid portability. This is often a raw object.

situate

A function. The situate() function is defined when bundle() is called, though is a loose analogue of an unbundle() S3 method for that object. Since the function is defined on bundle(), it has access to references and dependency information that can be saved alongside the object component. Calling unbundle() on a bundled object x calls x$situate(x$object), returning the unserialized version of object. situate() will also restore needed references, such as server instances and environmental variables.

Bundles are R objects that represent a "standalone" version of their analogous model object. Thus, bundles are ready for saving to a file; saving with base::saveRDS() is our recommended serialization strategy for bundles, unless documented otherwise for a specific method.

To restore the original model object x in a new environment, load its bundle with base::readRDS() and run unbundle() on it. The output of unbundle() is a model object that is ready to predict() on new data, and other restored functionality (like plotting or summarizing) is supported as a side effect only.

The bundle package wraps native serialization methods from model-supplying packages. Between versions, those model-supplying packages may change their native serialization methods, possibly introducing problems with re-loading objects serialized with previous package versions. The bundle package does not provide checks for these sorts of changes, and ought to be used in conjunction with tooling for managing and monitoring model environments like vetiver or renv.

See vignette("bundle") for more information on bundling and its motivation.

bundle: Serialize Model Objects With A Consistent Interface

Description

Typically, models in 'R' exist in memory and can be saved via regular 'R' serialization. However, some models store information in locations that cannot be saved using 'R' serialization alone. The goal of 'bundle' is to provide a common interface to capture this information, situate it within a portable object, and restore it for use in new settings.

Author(s)

Maintainer: Julia Silge [email protected] (ORCID)

Authors:

Simon Couch [email protected]
Qiushi Yan [email protected]
Max Kuhn [email protected]

Other contributors:

Posit Software, PBC [copyright holder, funder]

Bundle a `bart` object

Description

Bundling a model prepares it to be saved to a file and later restored for prediction in a new R session. See the 'Value' section for more information on bundles and their usage.

Usage

## S3 method for class 'bart'
bundle(x, ...)
## S3 method for class 'bart'
bundle(x, ...)

Arguments

`x`	A `bart` object returned from `dbarts::bart()`. Notably, this ought not to be the output of `parsnip::bart()`.
`...`	Not used in this bundler and included for compatibility with the generic only. Additional arguments passed to this method will return an error.

Value

A bundle object with subclass bundled_bart.

Bundles are a list subclass with two components:

object

An R object. Gives the output of native serialization methods from the model-supplying package, sometimes with additional classes or attributes that aid portability. This is often a raw object.

situate

See vignette("bundle") for more information on bundling and its motivation.

bundle and butcher

The butcher package allows you to remove parts of a fitted model object that are not needed for prediction.

This bundle method is compatible with pre-butchering. That is, for a fitted model x, you can safely call:

res <-
  x %>%
  butcher() %>%
  bundle()

and predict with the output of unbundle(res) in a new R session.

Examples


# fit model and bundle ------------------------------------------------
library(dbarts)

mtcars$vs <- as.factor(mtcars$vs)

set.seed(1)
fit <- dbarts::bart(mtcars[c("disp", "hp")], mtcars$vs, keeptrees = TRUE)

fit_bundle <- bundle(fit)

# then, after saveRDS + readRDS or passing to a new session ----------
fit_unbundled <- unbundle(fit_bundle)

fit_unbundled_preds <- predict(fit_unbundled, mtcars)

# fit model and bundle ------------------------------------------------
library(dbarts)

mtcars$vs <- as.factor(mtcars$vs)

set.seed(1)
fit <- dbarts::bart(mtcars[c("disp", "hp")], mtcars$vs, keeptrees = TRUE)

fit_bundle <- bundle(fit)

# then, after saveRDS + readRDS or passing to a new session ----------
fit_unbundled <- unbundle(fit_bundle)

fit_unbundled_preds <- predict(fit_unbundled, mtcars)

Bundle an `h2o` object

Description

Bundling a model prepares it to be saved to a file and later restored for prediction in a new R session. See the 'Value' section for more information on bundles and their usage.

Usage

## S3 method for class 'H2OAutoML'
bundle(x, id = NULL, n = NULL, ...)

## S3 method for class 'H2OMultinomialModel'
bundle(x, ...)

## S3 method for class 'H2OBinomialModel'
bundle(x, ...)

## S3 method for class 'H2ORegressionModel'
bundle(x, ...)
## S3 method for class 'H2OAutoML'
bundle(x, id = NULL, n = NULL, ...)

## S3 method for class 'H2OMultinomialModel'
bundle(x, ...)

## S3 method for class 'H2OBinomialModel'
bundle(x, ...)

## S3 method for class 'H2ORegressionModel'
bundle(x, ...)

Arguments

`x`	An object returned from modeling functions in the h2o package.
`id`	A single character. The `model_id` entry in the leaderboard. Applies to AutoML output only. Supply only one of this argument or `n`.
`n`	An integer giving the position in the leaderboard of the model to bundle. Applies to AutoML output only. Will be ignored if `id` is supplied.
`...`	Not used in this bundler and included for compatibility with the generic only. Additional arguments passed to this method will return an error.

Value

A bundle object with subclass bundled_h2o.

Bundles are a list subclass with two components:

object

An R object. Gives the output of native serialization methods from the model-supplying package, sometimes with additional classes or attributes that aid portability. This is often a raw object.

situate

See vignette("bundle") for more information on bundling and its motivation.

Examples


# fit model and bundle ------------------------------------------------
library(h2o)

set.seed(1)

h2o.init()

cars_h2o <- as.h2o(mtcars)

cars_fit <-
  h2o.glm(
    x = colnames(cars_h2o)[2:11],
    y = colnames(cars_h2o)[1],
    training_frame = cars_h2o
  )

cars_bundle <- bundle(cars_fit)

# then, after saveRDS + readRDS or passing to a new session ----------
cars_unbundled <- unbundle(cars_fit)

predict(cars_unbundled, cars_h2o[, 2:11])

h2o.shutdown(prompt = FALSE)

# fit model and bundle ------------------------------------------------
library(h2o)

set.seed(1)

h2o.init()

cars_h2o <- as.h2o(mtcars)

cars_fit <-
  h2o.glm(
    x = colnames(cars_h2o)[2:11],
    y = colnames(cars_h2o)[1],
    training_frame = cars_h2o
  )

cars_bundle <- bundle(cars_fit)

# then, after saveRDS + readRDS or passing to a new session ----------
cars_unbundled <- unbundle(cars_fit)

predict(cars_unbundled, cars_h2o[, 2:11])

h2o.shutdown(prompt = FALSE)

Bundle a `keras` object

Description

Bundling a model prepares it to be saved to a file and later restored for prediction in a new R session. See the 'Value' section for more information on bundles and their usage.

Usage

## S3 method for class 'keras.engine.training.Model'
bundle(x, ...)
## S3 method for class 'keras.engine.training.Model'
bundle(x, ...)

Arguments

`x`	An object returned from modeling functions in the keras package.
`...`	Not used in this bundler and included for compatibility with the generic only. Additional arguments passed to this method will return an error.

Details

This bundler does not currently support custom keras extensions, such as use of a keras::new_layer_class() or custom metric function. In such situations, consider using keras::with_custom_object_scope().

Value

A bundle object with subclass bundled_keras.

Bundles are a list subclass with two components:

object

An R object. Gives the output of native serialization methods from the model-supplying package, sometimes with additional classes or attributes that aid portability. This is often a raw object.

situate

See vignette("bundle") for more information on bundling and its motivation.

Examples


# fit model and bundle ------------------------------------------------
library(keras)

set.seed(1)

mnist <- dataset_mnist()
x_train <- mnist$train$x
y_train <- mnist$train$y
x_test <- mnist$test$x
y_test <- mnist$test$y

x_train <- array_reshape(x_train, c(nrow(x_train), 784))
x_test <- array_reshape(x_test, c(nrow(x_test), 784))

x_train <- x_train / 255
x_test <- x_test / 255

y_train <- to_categorical(y_train, 10)
y_test <- to_categorical(y_test, 10)

mod <- keras_model_sequential()

mod %>%
  layer_dense(units = 128, activation = 'relu', input_shape = c(784)) %>%
  layer_dropout(rate = 0.4) %>%
  layer_dense(units = 64, activation = 'relu') %>%
  layer_dropout(rate = 0.3) %>%
  layer_dense(units = 10, activation = 'softmax')

mod %>% compile(
  loss = 'categorical_crossentropy',
  optimizer = optimizer_rmsprop(),
  metrics = c('accuracy')
)

mod %>% fit(
  x_train, y_train,
  epochs = 5, batch_size = 128,
  validation_split = 0.2,
  verbose = 0
)

mod_bundle <- bundle(mod)

# then, after saveRDS + readRDS or passing to a new session ----------
mod_unbundled <- unbundle(mod_bundle)

predict(mod_unbundled, x_test)

# fit model and bundle ------------------------------------------------
library(keras)

set.seed(1)

mnist <- dataset_mnist()
x_train <- mnist$train$x
y_train <- mnist$train$y
x_test <- mnist$test$x
y_test <- mnist$test$y

x_train <- array_reshape(x_train, c(nrow(x_train), 784))
x_test <- array_reshape(x_test, c(nrow(x_test), 784))

x_train <- x_train / 255
x_test <- x_test / 255

y_train <- to_categorical(y_train, 10)
y_test <- to_categorical(y_test, 10)

mod <- keras_model_sequential()

mod %>%
  layer_dense(units = 128, activation = 'relu', input_shape = c(784)) %>%
  layer_dropout(rate = 0.4) %>%
  layer_dense(units = 64, activation = 'relu') %>%
  layer_dropout(rate = 0.3) %>%
  layer_dense(units = 10, activation = 'softmax')

mod %>% compile(
  loss = 'categorical_crossentropy',
  optimizer = optimizer_rmsprop(),
  metrics = c('accuracy')
)

mod %>% fit(
  x_train, y_train,
  epochs = 5, batch_size = 128,
  validation_split = 0.2,
  verbose = 0
)

mod_bundle <- bundle(mod)

# then, after saveRDS + readRDS or passing to a new session ----------
mod_unbundled <- unbundle(mod_bundle)

predict(mod_unbundled, x_test)

Bundle a `luz_module_fitted` object

Description

Bundling a model prepares it to be saved to a file and later restored for prediction in a new R session. See the 'Value' section for more information on bundles and their usage.

Usage

## S3 method for class 'luz_module_fitted'
bundle(x, ...)
## S3 method for class 'luz_module_fitted'
bundle(x, ...)

Arguments

`x`	A `luz_module_fitted` object returned from `luz::fit.luz_module_generator()`.
`...`	Not used in this bundler and included for compatibility with the generic only. Additional arguments passed to this method will return an error.

Details

For now, bundling methods for torch are only available via the luz package, "a higher level API for torch providing abstractions to allow for much less verbose training loops."

Value

A bundle object with subclass bundled_luz_module_fitted.

Bundles are a list subclass with two components:

object

An R object. Gives the output of native serialization methods from the model-supplying package, sometimes with additional classes or attributes that aid portability. This is often a raw object.

situate

See vignette("bundle") for more information on bundling and its motivation.

Examples


if (torch::torch_is_installed()) {
# fit model and bundle ------------------------------------------------
library(torch)
library(torchvision)
library(luz)

set.seed(1)

# example adapted from luz pkgdown article "Autoencoder"
dir <- tempdir()

mnist_dataset2 <- torch::dataset(
  inherit = mnist_dataset,
  .getitem = function(i) {
    output <- super$.getitem(i)
    output$y <- output$x
    output
  }
)

train_ds <- mnist_dataset2(
  dir,
  download = TRUE,
  transform = transform_to_tensor
)

test_ds <- mnist_dataset2(
  dir,
  train = FALSE,
  transform = transform_to_tensor
)

train_dl <- dataloader(train_ds, batch_size = 128, shuffle = TRUE)
test_dl <- dataloader(test_ds, batch_size = 128)

net <- nn_module(
  "Net",
  initialize = function() {
    self$encoder <- nn_sequential(
      nn_conv2d(1, 6, kernel_size=5),
      nn_relu(),
      nn_conv2d(6, 16, kernel_size=5),
      nn_relu()
    )
    self$decoder <- nn_sequential(
      nn_conv_transpose2d(16, 6, kernel_size = 5),
      nn_relu(),
      nn_conv_transpose2d(6, 1, kernel_size = 5),
      nn_sigmoid()
    )
  },
  forward = function(x) {
    x %>%
      self$encoder() %>%
      self$decoder()
  },
  predict = function(x) {
    self$encoder(x) %>%
      torch_flatten(start_dim = 2)
  }
)

mod <- net %>%
  setup(
    loss = nn_mse_loss(),
    optimizer = optim_adam
  ) %>%
  fit(train_dl, epochs = 1, valid_data = test_dl)

mod_bundle <- bundle(mod)


# then, after saveRDS + readRDS or passing to a new session ----------
mod_unbundled <- unbundle(mod_bundle)

mod_unbundled_preds <- predict(mod_unbundled, test_dl)
}

if (torch::torch_is_installed()) {
# fit model and bundle ------------------------------------------------
library(torch)
library(torchvision)
library(luz)

set.seed(1)

# example adapted from luz pkgdown article "Autoencoder"
dir <- tempdir()

mnist_dataset2 <- torch::dataset(
  inherit = mnist_dataset,
  .getitem = function(i) {
    output <- super$.getitem(i)
    output$y <- output$x
    output
  }
)

train_ds <- mnist_dataset2(
  dir,
  download = TRUE,
  transform = transform_to_tensor
)

test_ds <- mnist_dataset2(
  dir,
  train = FALSE,
  transform = transform_to_tensor
)

train_dl <- dataloader(train_ds, batch_size = 128, shuffle = TRUE)
test_dl <- dataloader(test_ds, batch_size = 128)

net <- nn_module(
  "Net",
  initialize = function() {
    self$encoder <- nn_sequential(
      nn_conv2d(1, 6, kernel_size=5),
      nn_relu(),
      nn_conv2d(6, 16, kernel_size=5),
      nn_relu()
    )
    self$decoder <- nn_sequential(
      nn_conv_transpose2d(16, 6, kernel_size = 5),
      nn_relu(),
      nn_conv_transpose2d(6, 1, kernel_size = 5),
      nn_sigmoid()
    )
  },
  forward = function(x) {
    x %>%
      self$encoder() %>%
      self$decoder()
  },
  predict = function(x) {
    self$encoder(x) %>%
      torch_flatten(start_dim = 2)
  }
)

mod <- net %>%
  setup(
    loss = nn_mse_loss(),
    optimizer = optim_adam
  ) %>%
  fit(train_dl, epochs = 1, valid_data = test_dl)

mod_bundle <- bundle(mod)


# then, after saveRDS + readRDS or passing to a new session ----------
mod_unbundled <- unbundle(mod_bundle)

mod_unbundled_preds <- predict(mod_unbundled, test_dl)
}

Bundle a parsnip `model_fit` object

Description

Bundling a model prepares it to be saved to a file and later restored for prediction in a new R session. See the 'Value' section for more information on bundles and their usage.

Usage

## S3 method for class 'model_fit'
bundle(x, ...)
## S3 method for class 'model_fit'
bundle(x, ...)

Arguments

`x`	A model_fit object returned from parsnip or other tidymodels packages.
`...`	Not used in this bundler and included for compatibility with the generic only. Additional arguments passed to this method will return an error.

Details

Primarily, these methods call bundle() on the output of parsnip::extract_fit_engine(). See the class of the output of that function for more details on the bundling method for that object.

Value

A bundle object with subclass bundled_model_fit.

Bundles are a list subclass with two components:

object

An R object. Gives the output of native serialization methods from the model-supplying package, sometimes with additional classes or attributes that aid portability. This is often a raw object.

situate

See vignette("bundle") for more information on bundling and its motivation.

bundle and butcher

The butcher package allows you to remove parts of a fitted model object that are not needed for prediction.

This bundle method is compatible with pre-butchering. That is, for a fitted model x, you can safely call:

res <-
  x %>%
  butcher() %>%
  bundle()

and predict with the output of unbundle(res) in a new R session.

Examples


# fit model and bundle ------------------------------------------------
library(parsnip)
library(xgboost)

set.seed(1)

mod <-
  boost_tree(trees = 5, mtry = 3) %>%
  set_mode("regression") %>%
  set_engine("xgboost") %>%
  fit(mpg ~ ., data = mtcars)

mod_bundle <- bundle(mod)

# then, after saveRDS + readRDS or passing to a new session ----------
mod_unbundled <- unbundle(mod_bundle)

mod_unbundled_preds <- predict(mod_unbundled, new_data = mtcars)

# fit model and bundle ------------------------------------------------
library(parsnip)
library(xgboost)

set.seed(1)

mod <-
  boost_tree(trees = 5, mtry = 3) %>%
  set_mode("regression") %>%
  set_engine("xgboost") %>%
  fit(mpg ~ ., data = mtcars)

mod_bundle <- bundle(mod)

# then, after saveRDS + readRDS or passing to a new session ----------
mod_unbundled <- unbundle(mod_bundle)

mod_unbundled_preds <- predict(mod_unbundled, new_data = mtcars)

Bundle a tidymodels `model_stack` object

Description

Bundling a model prepares it to be saved to a file and later restored for prediction in a new R session. See the 'Value' section for more information on bundles and their usage.

Usage

## S3 method for class 'model_stack'
bundle(x, ...)
## S3 method for class 'model_stack'
bundle(x, ...)

Arguments

`x`	A model_stack object returned from fit_members().
`...`	Not used in this bundler and included for compatibility with the generic only. Additional arguments passed to this method will return an error.

Details

This bundler wraps bundle.model_fit() and bundle.workflow(). Both the fitted members and the meta-learner (in x$coefs) are bundled.

Value

A bundle object with subclass bundled_model_stack.

Bundles are a list subclass with two components:

object

An R object. Gives the output of native serialization methods from the model-supplying package, sometimes with additional classes or attributes that aid portability. This is often a raw object.

situate

See vignette("bundle") for more information on bundling and its motivation.

Examples


# fit model and bundle ------------------------------------------------
library(stacks)

set.seed(1)

mod <-
  stacks() %>%
  add_candidates(reg_res_lr) %>%
  add_candidates(reg_res_svm) %>%
  blend_predictions(times = 10) %>%
  fit_members()

mod_bundle <- bundle(mod)

# then, after saveRDS + readRDS or passing to a new session ----------
mod_unbundled <- unbundle(mod_bundle)

# fit model and bundle ------------------------------------------------
library(stacks)

set.seed(1)

mod <-
  stacks() %>%
  add_candidates(reg_res_lr) %>%
  add_candidates(reg_res_svm) %>%
  blend_predictions(times = 10) %>%
  fit_members()

mod_bundle <- bundle(mod)

# then, after saveRDS + readRDS or passing to a new session ----------
mod_unbundled <- unbundle(mod_bundle)

Bundle a `recipe` object

Description

Bundling a model prepares it to be saved to a file and later restored for prediction in a new R session. See the 'Value' section for more information on bundles and their usage.

Usage

## S3 method for class 'recipe'
bundle(x, ...)
## S3 method for class 'recipe'
bundle(x, ...)

Arguments

`x`	A recipe object returned from recipes.
`...`	Not used in this bundler and included for compatibility with the generic only. Additional arguments passed to this method will return an error.

Details

The method call bundle() on every step in the recipe object. See the classes of individual steps for more details on the bundling method for that object.

Value

A bundle object with subclass bundled_recipe.

Bundles are a list subclass with two components:

object

An R object. Gives the output of native serialization methods from the model-supplying package, sometimes with additional classes or attributes that aid portability. This is often a raw object.

situate

See vignette("bundle") for more information on bundling and its motivation.

Bundle a `step_umap` object

Description

Bundling a model prepares it to be saved to a file and later restored for prediction in a new R session. See the 'Value' section for more information on bundles and their usage.

Usage

## S3 method for class 'step_umap'
bundle(x, ...)
## S3 method for class 'step_umap'
bundle(x, ...)

Arguments

`x`	A step_umap object returned from embed.
`...`	Not used in this bundler and included for compatibility with the generic only. Additional arguments passed to this method will return an error.

Value

A bundle object with subclass bundled_step_umap.

Bundles are a list subclass with two components:

object

An R object. Gives the output of native serialization methods from the model-supplying package, sometimes with additional classes or attributes that aid portability. This is often a raw object.

situate

See vignette("bundle") for more information on bundling and its motivation.

bundle and butcher

The butcher package allows you to remove parts of a fitted model object that are not needed for prediction.

This bundle method is compatible with pre-butchering. That is, for a fitted model x, you can safely call:

res <-
  x %>%
  butcher() %>%
  bundle()

and predict with the output of unbundle(res) in a new R session.

Examples


# fit model and bundle ------------------------------------------------
library(recipes)
library(embed)

set.seed(1)

rec <- recipe(Species ~ ., data = iris) %>%
  step_normalize(all_predictors()) %>%
  step_umap(all_predictors(), outcome = vars(Species), num_comp = 2) %>%
  prep()

rec_bundle <- bundle(rec)

# then, after saveRDS + readRDS or passing to a new session ----------
rec_unbundled <- unbundle(rec_bundle)

bake(rec_unbundled, new_data = iris)

# fit model and bundle ------------------------------------------------
library(recipes)
library(embed)

set.seed(1)

rec <- recipe(Species ~ ., data = iris) %>%
  step_normalize(all_predictors()) %>%
  step_umap(all_predictors(), outcome = vars(Species), num_comp = 2) %>%
  prep()

rec_bundle <- bundle(rec)

# then, after saveRDS + readRDS or passing to a new session ----------
rec_unbundled <- unbundle(rec_bundle)

bake(rec_unbundled, new_data = iris)

Bundle a caret `train` object

Description

Bundling a model prepares it to be saved to a file and later restored for prediction in a new R session. See the 'Value' section for more information on bundles and their usage.

Usage

## S3 method for class 'train'
bundle(x, ...)
## S3 method for class 'train'
bundle(x, ...)

Arguments

`x`	A train object returned from `caret::train()`.
`...`	Not used in this bundler and included for compatibility with the generic only. Additional arguments passed to this method will return an error.

Details

Primarily, these methods call bundle() on the output of train_model_object$finalModel. See the class of the output of that slot for more details on the bundling method for that object.

Value

A bundle object with subclass bundled_train.

Bundles are a list subclass with two components:

object

An R object. Gives the output of native serialization methods from the model-supplying package, sometimes with additional classes or attributes that aid portability. This is often a raw object.

situate

See vignette("bundle") for more information on bundling and its motivation.

bundle and butcher

The butcher package allows you to remove parts of a fitted model object that are not needed for prediction.

This bundle method is compatible with pre-butchering. That is, for a fitted model x, you can safely call:

res <-
  x %>%
  butcher() %>%
  bundle()

and predict with the output of unbundle(res) in a new R session.

Examples


# fit model and bundle ------------------------------------------------
library(caret)

predictors <- mtcars[, c("cyl", "disp", "hp")]

set.seed(1)

mod <-
  train(
    x = predictors,
    y = mtcars$mpg,
    method = "glm"
  )

mod_bundle <- bundle(mod)

# then, after saveRDS + readRDS or passing to a new session ----------
mod_unbundled <- unbundle(mod_bundle)

mod_unbundled_preds <- predict(mod_unbundled, new_data = mtcars)

# fit model and bundle ------------------------------------------------
library(caret)

predictors <- mtcars[, c("cyl", "disp", "hp")]

set.seed(1)

mod <-
  train(
    x = predictors,
    y = mtcars$mpg,
    method = "glm"
  )

mod_bundle <- bundle(mod)

# then, after saveRDS + readRDS or passing to a new session ----------
mod_unbundled <- unbundle(mod_bundle)

mod_unbundled_preds <- predict(mod_unbundled, new_data = mtcars)

Bundle a tidymodels `workflow` object

Description

Bundling a model prepares it to be saved to a file and later restored for prediction in a new R session. See the 'Value' section for more information on bundles and their usage.

Usage

## S3 method for class 'workflow'
bundle(x, ...)
## S3 method for class 'workflow'
bundle(x, ...)

Arguments

`x`	A workflow object returned from workflows or other tidymodels packages.
`...`	Not used in this bundler and included for compatibility with the generic only. Additional arguments passed to this method will return an error.

Details

This bundler wraps bundle.model_fit() and bundle.recipe().

Value

A bundle object with subclass bundled_workflow.

Bundles are a list subclass with two components:

object

An R object. Gives the output of native serialization methods from the model-supplying package, sometimes with additional classes or attributes that aid portability. This is often a raw object.

situate

See vignette("bundle") for more information on bundling and its motivation.

bundle and butcher

The butcher package allows you to remove parts of a fitted model object that are not needed for prediction.

This bundle method is compatible with pre-butchering. That is, for a fitted model x, you can safely call:

res <-
  x %>%
  butcher() %>%
  bundle()

and predict with the output of unbundle(res) in a new R session.

Examples


# fit model and bundle ------------------------------------------------
library(workflows)
library(recipes)
library(parsnip)
library(xgboost)

set.seed(1)

spec <-
  boost_tree(trees = 5, mtry = 3) %>%
  set_mode("regression") %>%
  set_engine("xgboost")

rec <-
  recipe(mpg ~ ., data = mtcars) %>%
  step_log(hp)

mod <-
  workflow() %>%
  add_model(spec) %>%
  add_recipe(rec) %>%
  fit(data = mtcars)

mod_bundle <- bundle(mod)

# then, after saveRDS + readRDS or passing to a new session ----------
mod_unbundled <- unbundle(mod_bundle)

# fit model and bundle ------------------------------------------------
library(workflows)
library(recipes)
library(parsnip)
library(xgboost)

set.seed(1)

spec <-
  boost_tree(trees = 5, mtry = 3) %>%
  set_mode("regression") %>%
  set_engine("xgboost")

rec <-
  recipe(mpg ~ ., data = mtcars) %>%
  step_log(hp)

mod <-
  workflow() %>%
  add_model(spec) %>%
  add_recipe(rec) %>%
  fit(data = mtcars)

mod_bundle <- bundle(mod)

# then, after saveRDS + readRDS or passing to a new session ----------
mod_unbundled <- unbundle(mod_bundle)

Bundle an `xgb.Booster` object

Description

Bundling a model prepares it to be saved to a file and later restored for prediction in a new R session. See the 'Value' section for more information on bundles and their usage.

Usage

## S3 method for class 'xgb.Booster'
bundle(x, ...)
## S3 method for class 'xgb.Booster'
bundle(x, ...)

Arguments

`x`	An `xgb.Booster` object returned from `xgboost::xgboost()` or `xgboost::xgb.train()`.
`...`	Not used in this bundler and included for compatibility with the generic only. Additional arguments passed to this method will return an error.

Value

A bundle object with subclass bundled_xgb.Booster.

Bundles are a list subclass with two components:

object

An R object. Gives the output of native serialization methods from the model-supplying package, sometimes with additional classes or attributes that aid portability. This is often a raw object.

situate

See vignette("bundle") for more information on bundling and its motivation.

bundle and butcher

The butcher package allows you to remove parts of a fitted model object that are not needed for prediction.

This bundle method is compatible with pre-butchering. That is, for a fitted model x, you can safely call:

res <-
  x %>%
  butcher() %>%
  bundle()

and predict with the output of unbundle(res) in a new R session.

Examples


# fit model and bundle ------------------------------------------------
library(xgboost)

set.seed(1)

data(agaricus.train)
data(agaricus.test)

xgb <- xgboost(data = agaricus.train$data, label = agaricus.train$label,
               max_depth = 2, eta = 1, nthread = 2, nrounds = 2,
               objective = "binary:logistic")

xgb_bundle <- bundle(xgb)

# then, after saveRDS + readRDS or passing to a new session ----------
xgb_unbundled <- unbundle(xgb_bundle)

xgb_unbundled_preds <- predict(xgb_unbundled, agaricus.test$data)

# fit model and bundle ------------------------------------------------
library(xgboost)

set.seed(1)

data(agaricus.train)
data(agaricus.test)

xgb <- xgboost(data = agaricus.train$data, label = agaricus.train$label,
               max_depth = 2, eta = 1, nthread = 2, nrounds = 2,
               objective = "binary:logistic")

xgb_bundle <- bundle(xgb)

# then, after saveRDS + readRDS or passing to a new session ----------
xgb_unbundled <- unbundle(xgb_bundle)

xgb_unbundled_preds <- predict(xgb_unbundled, agaricus.test$data)

Package 'bundle'

Help Index

Bundling

Description

Usage

Arguments

Value

See Also

bundle: Serialize Model Objects With A Consistent Interface

Description

Author(s)

See Also

Bundle a bart object

Description

Usage

Arguments

Value

bundle and butcher

See Also

Examples

Bundle an h2o object

Description

Usage

Arguments

Value

See Also

Examples

Bundle a keras object

Description

Usage

Arguments

Details

Value

See Also

Examples

Bundle a luz_module_fitted object

Description

Usage

Arguments

Details

Value

See Also

Examples

Bundle a parsnip model_fit object

Description

Usage

Arguments

Details

Value

bundle and butcher

See Also

Examples

Bundle a tidymodels model_stack object

Description

Usage

Arguments

Details

Value

See Also

Examples

Bundle a recipe object

Description

Usage

Arguments

Details

Value

See Also

Bundle a step_umap object

Description

Usage

Arguments

Value

bundle and butcher

See Also

Examples

Bundle a caret train object

Description

Usage

Arguments

Details

Bundle a `bart` object

Bundle an `h2o` object

Bundle a `keras` object

Bundle a `luz_module_fitted` object

Bundle a parsnip `model_fit` object

Bundle a tidymodels `model_stack` object

Bundle a `recipe` object

Bundle a `step_umap` object

Bundle a caret `train` object

Bundle a tidymodels `workflow` object

Bundle an `xgb.Booster` object