Title:	More Miscellaneous Steps for the 'recipes' Package
Version:	0.1.0
Description:	Contains additional miscellaneous steps for the 'recipes' package. These steps are useful, but doesn't have a good home in other 'recipes' packages or its extensions.
License:	MIT + file LICENSE
URL:	https://github.com/EmilHvitfeldt/extrasteps, https://emilhvitfeldt.github.io/extrasteps/
BugReports:	https://github.com/EmilHvitfeldt/extrasteps/issues
Depends:	R (≥ 3.6), recipes (≥ 1.0.7)
Imports:	dplyr, generics, magrittr, purrr, rlang, tibble, vctrs
Suggests:	almanac, ggplot2, modeldata, testthat (≥ 3.0.0)
Config/testthat/edition:	3
Encoding:	UTF-8
RoxygenNote:	7.3.2
NeedsCompilation:	no
Packaged:	2024-10-02 18:04:01 UTC; emilhvitfeldt
Author:	Emil Hvitfeldt [aut, cre]
Maintainer:	Emil Hvitfeldt <emilhhvitfeldt@gmail.com>
Repository:	CRAN
Date/Publication:	2024-10-03 19:20:02 UTC

extrasteps: More Miscellaneous Steps for the 'recipes' Package

Description

Contains additional miscellaneous steps for the 'recipes' package. These steps are useful, but doesn't have a good home in other 'recipes' packages or its extensions.

Author(s)

Maintainer: Emil Hvitfeldt emilhhvitfeldt@gmail.com (ORCID)

See Also

Useful links:

• https://github.com/EmilHvitfeldt/extrasteps

• https://emilhvitfeldt.github.io/extrasteps/

• Report bugs at https://github.com/EmilHvitfeldt/extrasteps/issues

Objects exported from other packages

Description

These objects are imported from other packages. Follow the links below to see their documentation.

generics

required_pkgs

magrittr

%>%

S3 methods for tracking which additional packages are needed for steps.

Description

Recipe-adjacent packages always list themselves as a required package so that the steps can function properly within parallel processing schemes.

Usage

## S3 method for class 'step_date_after'
required_pkgs(x, ...)

## S3 method for class 'step_date_before'
required_pkgs(x, ...)

## S3 method for class 'step_date_nearest'
required_pkgs(x, ...)

## S3 method for class 'step_difftime'
required_pkgs(x, ...)

## S3 method for class 'step_encoding_binary'
required_pkgs(x, ...)

## S3 method for class 'step_encoding_frequency'
required_pkgs(x, ...)

## S3 method for class 'step_maxabs'
required_pkgs(x, ...)

## S3 method for class 'step_minmax'
required_pkgs(x, ...)

## S3 method for class 'step_robust'
required_pkgs(x, ...)

## S3 method for class 'step_time_event'
required_pkgs(x, ...)

## S3 method for class 'step_unit_normalize'
required_pkgs(x, ...)

Arguments

Value

A character vector

Time after Recurrent Date Time Event

Description

step_date_after() creates a specification of a recipe step that will create new columns indicating the time after an recurrent event.

Usage

step_date_after(
  recipe,
  ...,
  role = "predictor",
  trained = FALSE,
  rules = list(),
  transform = "identity",
  columns = NULL,
  skip = FALSE,
  id = rand_id("date_after")
)

Arguments

Details

The transform argument can be function that takes a numeric vector and returns a numeric vector of the same length. It can also be a character vector, below is the supported vector names. Some functions come with offset to avoid Inf.

"identity"
function(x) x

"inverse"
function(x) 1 / (x + 0.5)

"exp"
function(x) exp(x)

"log"
function(x) log(x + 0.5)

The effect of transform is illustrated below.

The naming of the resulting variables will be on the form

{variable name}_after_{name of rule}

Value

An updated version of recipe with the new check added to the sequence of any existing operations.

Examples

library(recipes)
library(extrasteps)
library(almanac)
library(modeldata)

data(Chicago)

on_easter <- yearly() %>% recur_on_easter()
on_weekend <- weekly() %>% recur_on_weekends()

rules <- list(easter = on_easter, weekend = on_weekend)

rec_spec <- recipe(ridership ~ date, data = Chicago) %>%
  step_date_after(date, rules = rules)

rec_spec_preped <- prep(rec_spec)

bake(rec_spec_preped, new_data = NULL)

Time before Recurrent Date Time Event

Description

step_date_before() creates a specification of a recipe step that will create new columns indicating the time before an recurrent event.

Usage

step_date_before(
  recipe,
  ...,
  role = "predictor",
  trained = FALSE,
  rules = list(),
  transform = "identity",
  columns = NULL,
  skip = FALSE,
  id = rand_id("date_before")
)

Arguments

Details

The transform argument can be function that takes a numeric vector and returns a numeric vector of the same length. It can also be a character vector, below is the supported vector names. Some functions come with offset to avoid Inf.

"identity"
function(x) x

"inverse"
function(x) 1 / (x + 0.5)

"exp"
function(x) exp(x)

"log"
function(x) log(x + 0.5)

The effect of transform is illustrated below.

The naming of the resulting variables will be on the form

{variable name}_before_{name of rule}

Value

An updated version of recipe with the new check added to the sequence of any existing operations.

Examples

library(recipes)
library(extrasteps)
library(almanac)
library(modeldata)

data(Chicago)

on_easter <- yearly() %>% recur_on_easter()
on_weekend <- weekly() %>% recur_on_weekends()

rules <- list(easter = on_easter, weekend = on_weekend)

rec_spec <- recipe(ridership ~ date, data = Chicago) %>%
  step_date_before(date, rules = rules)

rec_spec_preped <- prep(rec_spec)

bake(rec_spec_preped, new_data = NULL)

Time to Nearest Recurrent Date Time Event

Description

step_date_nearest() creates a specification of a recipe step that will create new columns indicating the time to nearest recurrent event.

Usage

step_date_nearest(
  recipe,
  ...,
  role = "predictor",
  trained = FALSE,
  rules = list(),
  transform = "identity",
  columns = NULL,
  skip = FALSE,
  id = rand_id("date_nearest")
)

Arguments

Details

The transform argument can be function that takes a numeric vector and returns a numeric vector of the same length. It can also be a character vector, below is the supported vector names. Some functions come with offset to avoid Inf.

"identity"
function(x) x

"inverse"
function(x) 1 / (x + 0.5)

"exp"
function(x) exp(x)

"log"
function(x) log(x + 0.5)

The effect of transform is illustrated below.

The naming of the resulting variables will be on the form

{variable name}_nearest_{name of rule}

Value

An updated version of recipe with the new check added to the sequence of any existing operations.

Examples

library(recipes)
library(extrasteps)
library(almanac)
library(modeldata)

data(Chicago)

on_easter <- yearly() %>% recur_on_easter()
on_weekend <- weekly() %>% recur_on_weekends()

rules <- list(easter = on_easter, weekend = on_weekend)

rec_spec <- recipe(ridership ~ date, data = Chicago) %>%
  step_date_nearest(date, rules = rules)

rec_spec_preped <- prep(rec_spec)

bake(rec_spec_preped, new_data = NULL)

difftimearithmic Transformation

Description

step_difftime() creates a specification of a recipe step that will calculate difftimes of the data.

Usage

step_difftime(
  recipe,
  ...,
  role = NA,
  trained = FALSE,
  time = NULL,
  tz = NULL,
  unit = "auto",
  columns = NULL,
  skip = FALSE,
  id = rand_id("difftime")
)

Arguments

Value

An updated version of recipe with the new step added to the sequence of existing steps (if any). For the tidy method, a tibble with columns terms (the columns that will be affected) and base.

Examples

library(recipes)
example_date <- data.frame(
  dates = seq(as.Date("2010/1/1"), as.Date("2016/1/1"), by = "quarter")
 )

 example_datetime <- data.frame(
  datetimes = seq(ISOdate(1993,1,1), ISOdate(1993,1,2), by = "hour")
 )

rec <- recipe(~ dates, data = example_date) %>%
  step_difftime(dates, time = as.Date("2010/1/1"))

difftime_obj <- prep(rec)

bake(difftime_obj, new_data = NULL)

recipe(~ dates, data = example_date) %>%
  step_difftime(dates, time = as.Date("2010/1/1"), unit = "weeks") %>%
  prep() %>%
  bake(new_data = NULL)

recipe(~ datetimes, data = example_datetime) %>%
  step_difftime(datetimes, time = ISOdate(1993,1,1), unit = "secs") %>%
  prep() %>%
  bake(new_data = NULL)

Perform binary encoding of factor variables

Description

step_encoding_binary() creates a specification of a recipe step that will perform binary encoding of factor variables.

Usage

step_encoding_binary(
  recipe,
  ...,
  role = NA,
  trained = FALSE,
  res = NULL,
  columns = NULL,
  keep_original_cols = FALSE,
  skip = FALSE,
  id = rand_id("encoding_binary")
)

Arguments

Value

An updated version of recipe with the new step added to the sequence of existing steps (if any). For the tidy method, a tibble with columns terms (the columns that will be affected) and base.

Examples

library(recipes)
library(modeldata)

data(ames)

rec <- recipe(~ Land_Contour + Neighborhood, data = ames) %>%
  step_encoding_binary(all_nominal_predictors()) %>%
  prep()

rec %>%
  bake(new_data = NULL)

tidy(rec, 1)

Perform frequency encoding

Description

step_encoding_frequency() creates a specification of a recipe step that will perform frequency encoding.

Usage

step_encoding_frequency(
  recipe,
  ...,
  role = NA,
  trained = FALSE,
  res = NULL,
  columns = NULL,
  skip = FALSE,
  id = rand_id("encoding_frequency")
)

Arguments

Value

An updated version of recipe with the new step added to the sequence of existing steps (if any). For the tidy method, a tibble with columns terms (the columns that will be affected) and base.

Examples

library(recipes)
library(modeldata)

data(ames)

rec <- recipe(~ Land_Contour + Neighborhood, data = ames) %>%
  step_encoding_frequency(all_nominal_predictors()) %>%
  prep()

rec %>%
  bake(new_data = NULL)

tidy(rec, 1)

Perform Max Abs Scaling

Description

step_maxabs() creates a specification of a recipe step that will perform Max Abs scaling.

Usage

step_maxabs(
  recipe,
  ...,
  role = NA,
  trained = FALSE,
  res = NULL,
  columns = NULL,
  skip = FALSE,
  id = rand_id("maxabs")
)

Arguments

Value

An updated version of recipe with the new step added to the sequence of existing steps (if any). For the tidy method, a tibble with columns terms (the columns that will be affected) and base.

Examples

library(recipes)

rec <- recipe(~., data = mtcars) %>%
  step_maxabs(all_predictors()) %>%
  prep()

rec %>%
  bake(new_data = NULL)

tidy(rec, 1)

Perform Min Max Scaling

Description

step_minmax() creates a specification of a recipe step that will perform Min Max scaling.

Usage

step_minmax(
  recipe,
  ...,
  role = NA,
  trained = FALSE,
  res = NULL,
  columns = NULL,
  skip = FALSE,
  id = rand_id("minmax")
)

Arguments

Value

An updated version of recipe with the new step added to the sequence of existing steps (if any). For the tidy method, a tibble with columns terms (the columns that will be affected) and base.

Examples

library(recipes)

rec <- recipe(~., data = mtcars) %>%
  step_minmax(all_predictors()) %>%
  prep()

rec %>%
  bake(new_data = NULL)

tidy(rec, 1)

Perform Robust Scaling

Description

step_robust() creates a specification of a recipe step that will perform Robust scaling.

Usage

step_robust(
  recipe,
  ...,
  role = NA,
  trained = FALSE,
  range = c(0.25, 0.75),
  res = NULL,
  columns = NULL,
  skip = FALSE,
  id = rand_id("robust")
)

Arguments

Details

The scaling performed by this step is done using the following transformation

x_new = (x - Q2(x)) / (Q3(x) - Q1(x))

where Q2(x) is the median, Q3(x) is the upper quantile (defaults to 0.75) and Q1(x) is the lower quantile (defaults to 0.25). The upper and lower quantiles can be changed with the range argument.

Value

An updated version of recipe with the new step added to the sequence of existing steps (if any). For the tidy method, a tibble with columns terms (the columns that will be affected) and base.

Examples

library(recipes)

rec <- recipe(~., data = mtcars) %>%
  step_robust(all_predictors()) %>%
  prep()

rec %>%
  bake(new_data = NULL)

tidy(rec, 1)

rec <- recipe(~., data = mtcars) %>%
  step_robust(all_predictors(), range = c(0.1, 0.9)) %>%
  prep()

rec %>%
  bake(new_data = NULL)

tidy(rec, 1)

Indicate Recurrent Date Time Event

Description

step_time_event() creates a specification of a recipe step that will create new columns indicating if the date fall on recurrent event.

Usage

step_time_event(
  recipe,
  ...,
  role = "predictor",
  trained = FALSE,
  rules = list(),
  columns = NULL,
  keep_original_cols = FALSE,
  skip = FALSE,
  id = rand_id("time_event")
)

Arguments

Details

Unlike some other steps step_time_event does not remove the original date variables by default. Set keep_original_cols to FALSE to remove them.

Value

An updated version of recipe with the new check added to the sequence of any existing operations.

Examples

library(recipes)
library(extrasteps)
library(almanac)
library(modeldata)

data(Chicago)

on_easter <- yearly() %>% recur_on_easter()
on_weekend <- weekly() %>% recur_on_weekends()

rules <- list(easter = on_easter, weekend = on_weekend)

rec_spec <- recipe(ridership ~ date, data = Chicago) %>%
  step_time_event(date, rules = rules)

rec_spec_preped <- prep(rec_spec)

bake(rec_spec_preped, new_data = NULL)

Perform Unit Normalization

Description

step_unit_normalize() creates a specification of a recipe step that will perform unit normalization by scaling individual samples to have unit norm.

Usage

step_unit_normalize(
  recipe,
  ...,
  role = NA,
  trained = FALSE,
  norm = c("l2", "l1", "max"),
  columns = NULL,
  skip = FALSE,
  id = rand_id("unit_normalize")
)

Arguments

Value

An updated version of recipe with the new step added to the sequence of existing steps (if any). For the tidy method, a tibble with columns terms (the columns that will be affected) and base.

Examples

library(recipes)

rec <- recipe(~., data = mtcars) %>%
  step_unit_normalize(all_predictors()) %>%
  prep()

rec %>%
  bake(new_data = NULL)

tidy(rec, 1)