Calculate quantiles with grouping for multiple columns with dplyr-CodePudding

Tidyverse provides an example of how to calculate quantiles with grouping for a single column, x. Below, I've modified that example to include column z which is character and a missing value (e.g., so needing to use na.rm=TRUE).

I want to modify the below modified code so that works it works for all numeric columns (i.e., is.numeric)

library(tidyverse)
df <- tibble(
  grp = rep(c("A","B"), each = 5), 
  x = c(rnorm(5, -0.25, 1), rnorm(5, 0, 1.5)),
  y = c(rnorm(5, 0.25, 1), rnorm(5, 0, 0.5)),
  z = letters[1:10],
)

df[3,2] <-NA
df

# A tibble: 10 × 4
   grp         x       y z    
   <chr>   <dbl>   <dbl> <chr>
 1 A      1.02   -0.653  a    
 2 A     -0.370  -0.185  b    
 3 A     NA      -1.49   c    
 4 A     -0.336   0.408  d    
 5 A     -0.0526  1.90   e    
 6 B      1.45   -0.0769 f    
 7 B      1.36   -1.11   g    
 8 B      0.0972 -0.0394 h    
 9 B     -1.72   -0.198  i    
10 B     -2.50    0.364  j 

quibble <- function(x, q = c(0.25, 0.5, 0.75), dropNA = TRUE) {
  tibble(x = quantile(x, q, na.rm = dropNA), q = q)
}

df %>% 
  group_by(grp) %>% 
  summarise(x = list(quibble(x, c(0.25, 0.75), dropNA = TRUE))) %>% 
  tidyr::unnest(x)

# A tibble: 4 × 3
  grp        x     q
  <chr>  <dbl> <dbl>
1 A     -0.345  0.25
2 A      0.215  0.75
3 B     -1.72   0.25
4 B      1.36   0.75

My attempt has included the below which was inspired from here

df %>% 
  group_by(grp) %>%
  summarise(across(everything(), where(is.numeric) ~ quibble(.x, c(0.25, 0.75), dropNA = TRUE))) %>% 
  tidyr::unnest(.x)

I'm aware of this page which includes sapply options in comments and a solution that I have not been able to apply to this situation (multiple numeric and columns, missing values, and grouping).

CodePudding user response：

Here, we don't need both everything() and is.numeric. In addition, there should be a , between the where expression and the lambda expression

library(dplyr)
library(tidyr)
df %>% 
  group_by(grp) %>%
  summarise(across( where(is.numeric), ~ quibble(.x, c(0.25, 0.75), 
     dropNA = TRUE)), .groups = 'drop') %>% 
  tidyr::unnest(where(is_tibble), names_repair = 'unique', names_sep = "_")

-output

# A tibble: 4 × 5
  grp        x_x   x_q    y_x   y_q
  <chr>    <dbl> <dbl>  <dbl> <dbl>
1 A      0.00519  0.25 -0.888  0.25
2 A      0.723    0.75  1.09   0.75
3 B     -1.54     0.25  0.213  0.25
4 B     -0.938    0.75  0.439  0.75