Skip to content

tidymodels/textrecipes

Repository files navigation

textrecipes

R-CMD-check Codecov test coverage CRAN status Downloads Lifecycle: maturing

Introduction

textrecipes contain extra steps for the recipes package for preprocessing text data.

Installation

You can install the released version of textrecipes from CRAN with:

install.packages("textrecipes")

Install the development version from GitHub with:

# install.packages("pak")
pak::pak("tidymodels/textrecipes")

Example

In the following example we will go through the steps needed, to convert a character variable to the TF-IDF of its tokenized words after removing stopwords, and, limiting ourself to only the 10 most used words. The preprocessing will be conducted on the variable medium and artist.

library(recipes)
library(textrecipes)
library(modeldata)

data("tate_text")

okc_rec <- recipe(~ medium + artist, data = tate_text) %>%
  step_tokenize(medium, artist) %>%
  step_stopwords(medium, artist) %>%
  step_tokenfilter(medium, artist, max_tokens = 10) %>%
  step_tfidf(medium, artist)

okc_obj <- okc_rec %>%
  prep()

str(bake(okc_obj, tate_text))
#> tibble [4,284 × 20] (S3: tbl_df/tbl/data.frame)
#>  $ tfidf_medium_colour     : num [1:4284] 2.31 0 0 0 0 ...
#>  $ tfidf_medium_etching    : num [1:4284] 0 0.86 0.86 0.86 0 ...
#>  $ tfidf_medium_gelatin    : num [1:4284] 0 0 0 0 0 0 0 0 0 0 ...
#>  $ tfidf_medium_lithograph : num [1:4284] 0 0 0 0 0 0 0 0 0 0 ...
#>  $ tfidf_medium_paint      : num [1:4284] 0 0 0 0 2.35 ...
#>  $ tfidf_medium_paper      : num [1:4284] 0 0.422 0.422 0.422 0 ...
#>  $ tfidf_medium_photograph : num [1:4284] 0 0 0 0 0 0 0 0 0 0 ...
#>  $ tfidf_medium_print      : num [1:4284] 0 0 0 0 0 ...
#>  $ tfidf_medium_screenprint: num [1:4284] 0 0 0 0 0 0 0 0 0 0 ...
#>  $ tfidf_medium_silver     : num [1:4284] 0 0 0 0 0 0 0 0 0 0 ...
#>  $ tfidf_artist_akram      : num [1:4284] 0 0 0 0 0 0 0 0 0 0 ...
#>  $ tfidf_artist_beuys      : num [1:4284] 0 0 0 0 0 ...
#>  $ tfidf_artist_ferrari    : num [1:4284] 0 0 0 0 0 0 0 0 0 0 ...
#>  $ tfidf_artist_john       : num [1:4284] 0 0 0 0 0 0 0 0 0 0 ...
#>  $ tfidf_artist_joseph     : num [1:4284] 0 0 0 0 0 ...
#>  $ tfidf_artist_león       : num [1:4284] 0 0 0 0 0 0 0 0 0 0 ...
#>  $ tfidf_artist_richard    : num [1:4284] 0 0 0 0 0 0 0 0 0 0 ...
#>  $ tfidf_artist_schütte    : num [1:4284] 0 0 0 0 0 0 0 0 0 0 ...
#>  $ tfidf_artist_thomas     : num [1:4284] 0 0 0 0 0 0 0 0 0 0 ...
#>  $ tfidf_artist_zaatari    : num [1:4284] 0 0 0 0 0 0 0 0 0 0 ...

Breaking changes

As of version 0.4.0, step_lda() no longer accepts character variables and instead takes tokenlist variables.

the following recipe

recipe(~text_var, data = data) %>%
  step_lda(text_var)

can be replaced with the following recipe to achive the same results

lda_tokenizer <- function(x) text2vec::word_tokenizer(tolower(x))
recipe(~text_var, data = data) %>%
  step_tokenize(text_var,
    custom_token = lda_tokenizer
  ) %>%
  step_lda(text_var)

Contributing

This project is released with a Contributor Code of Conduct. By contributing to this project, you agree to abide by its terms.