Use function over groups or factors with dplyr

Question

I want to use a function, e.g. shapiro.test() over several groups in a dataset.

First I tried

library(tidyverse)
library(magrittr)

mtcars %>% group_by(cyl) %$% shapiro.test(wt)$p.value
#> [1] 0.09265499

But that did not iterate over the groups as I expected. Then I tried a function that would output the results as a dataframe, as that was the approach taken for another question here on Stack Overflow.

checkNorm <- function(x) {
  return(data.frame(P = shapiro.test(x)$p.value))
}

mtcars %>% group_by(cyl) %$% checkNorm(wt)
#>            P
#> 1 0.09265499

What is the appropriate way to make functions iterate over the groups passed by group_by()?

Answer 1

Create a new column to store p-value of each group :

library(dplyr)

mtcars %>% 
  group_by(cyl) %>%
  summarise(p_val = shapiro.test(wt)$p.value)

#   cyl   p_val
#  <dbl>   <dbl>
#1     4 0.570  
#2     6 0.131  
#3     8 0.00275