Apply boxplot function within nests-CodePudding

I have a simple nest that groups by colour for iris then I want to apply a boxplot and grab the outliers from the dataset after it has been grouped.

For example:

bxplt <- function(DATA, X, Y){
  bxp <- boxplot(data = DATA, {{Y}}~{{X}}, outcex=2)
  cBXP <- bxp[4:5] %>% do.call(cbind.data.frame, bxp)
  return(xBXP)
}
upIris <- iris %>% mutate(colour = rep(c('Red', 'Blue', 'Green'), 50))
upIris %>% group_by(colour) %>% nest(.) %>% mutate(extra_data = map(data, ~ .x %>% bxplt(., Species, Sepal.Length)))

This gives the following error:

Error in mutate(): ! Problem while computing extra_data = map(data, ~.x %>% bxplt(., Species, Sepal.Length)). ℹ The error occurred in group 1: colour = "Blue". Caused by error: ! object 'Sepal.Length' not found

CodePudding user response：

You can use ensym() and inject() from rlang

library(rlang)
bxplt <- function(DATA, X, Y){
  y <- ensym(Y); x <- ensym(X)
  bxp <- inject(boxplot(!!y ~ !!x,data = DATA, outcex=2))
  do.call(cbind.data.frame, bxp[4:5])
}

Now, apply that function using map():

iris %>%
  mutate(colour = rep(c('Red', 'Blue', 'Green'), 50)) %>% 
  group_by(colour) %>% 
  nest() %>% 
  mutate(extra_data = map(data, ~bxplt(.x, Species, Sepal.Length)))

Output:

  colour data              extra_data  
  <chr>  <list>            <list>      
1 Red    <tibble [50 x 5]> <df [0 x 2]>
2 Blue   <tibble [50 x 5]> <df [8 x 2]>
3 Green  <tibble [50 x 5]> <df [1 x 2]>