Using purrr for rowwise operations

Question

I'm looking to use purrr for a rowwise operation where each element is used in the function as a string. The error I'm getting is that it can't attribute an $ to an atomic vector. Here's an example:

test_function = function(dat_) {
  
  petal_width = dat_$Petal.Width
  sepal_width = dat_$Sepal.Width
  petal_length = dat_$Petal.Length
  sepal_length = dat_$Sepal.Length
  
  list(petal_length,sepal_length,sepal_width,petal_width) %>%
    bind_cols -> test
  
  return(test)
  
}

apply(iris, 1, test_function)

Answer 1

Not sure what your expected output is but you can use pmap to perform row-wise operations. However, pmap passes each row as a vector and not as dataframe so $ will not work. You can change the function to :

library(tidyverse)

test_function = function(dat_) {

  petal_width = dat_[['Petal.Width']]
  sepal_width = dat_[['Sepal.Width']]
  petal_length = dat_[['Petal.Length']]
  sepal_length = dat_[['Sepal.Length']]
  tibble(a = petal_length,b = sepal_length,
         c = sepal_width,d = petal_width) -> test
  return(test)
}

You can use pmap as :

iris %>% mutate(data = pmap(select(., matches('Sepal|Petal')), 
               ~test_function(c(...)))) -> tmp

Each row of data is a tibble with data from that row.

tmp$data[[1]]
# A tibble: 1 x 4
      a     b     c     d
  <dbl> <dbl> <dbl> <dbl>
1   1.4   5.1   3.5   0.2