R tidymodels recipes step - from a column with multiple values, create a new column for each of these values (one hot encoding)

Question

Say i have this dataframe:

library(tidyverse)
# Sample data frame
df <- data.frame(
  id = 1:3,
  fruits = c("apple | oranges", "apple | bananas", "bananas | oranges")
)
df

id	fruits
1	apple | oranges
2	apple | bananas
3	bananas | oranges

I want to separate the values from the fruits column and then perform one hot encoding for each one as follows:

# Step 1: Separate the values based on |
df_separated <- df %>%
  separate_rows(fruits, sep = " \\| ")

# Step 2: Create a dummy variable for each element
df_dummy <- df_separated %>%
  mutate(value = TRUE) %>%
  spread(fruits, value, fill = FALSE)

# View the result
print(df_dummy)

id	apple	bananas	oranges
1	TRUE	FALSE	TRUE
2	TRUE	TRUE	FALSE
3	FALSE	TRUE	TRUE

However, I cannot manage to convert this code as a recipe step to incorporate it into a tidymodels workflow. Any ideas how to do so?

Answer 1

library(tidymodels)

dummies_fruit <- recipe(~ fruits, data = df) |>
  step_dummy_extract(fruits, sep = " | ") |>
  prep()