CODEWITHSUNDEEP
rdplyr
b-ryce
b-ryce

Reputation: 5828

Conditional case inside dplyr

I'm trying to conditionally reorder columns but only when when specified by a function parameter. I see other examples that show how to add conditionals inside dplyr statements,(like this) but it isn't working for me.

I've got a dataset with two columns that looks like this:

Role
1 Primary
2 Secondary
3 Primary
4 Primary

Zap
1 Enabled
2 Disabled
3 Enabled
4 Enabled

And my code looks like this:

my_bar_chart <- function(data, column, title, change_order=FALSE){
  toReturn <- data %>% 
    group_by({{column}}) %>% 
    summarize(count = n()) %>% 
    mutate(percent = count/sum(count)) %>% 
    mutate(column = reorder({{column}}, -count, FUN=identity)) %>%
    ggplot(aes(x=column, y=count)) +
    xlab(title)+
    geom_col(fill='#cccccc') +
    geom_text(aes(label = paste0(round(100 * percent, 1), "%")))
  return(toReturn)
}
p1 <- my_bar_chart(mydata, Role, "EA5 Controller Role", TRUE)
p2 <- my_bar_chart(mydata, Zap, "Another title", FALSE)

And that works ok, but now when I try to get working is the second mutate statement, to conditionally reorder when I pass in TRUE into the function change_order param. Here is the conditional mutate that isn't working:

mutate(column = case_when(
      change_order == TRUE ~ reorder({{column}}, -count, FUN=identity),
      change_order == FALSE ~ column)) %>%

It is giving me this error:

Error in eval_tidy(pair$rhs, env = default_env) : object 'Role' not found

Any ideas what I'm doing wrong?

Upvotes: 1

Views: 131

Answers (1)

akrun
akrun

Reputation: 886938

The second condition is returning column and not inside the {{}}, along with the type of object that should be the same 

my_bar_chart <- function(data, column, title, change_order=FALSE){
  data %>% 
    group_by({{column}}) %>% 
    summarize(count = n()) %>% 
    mutate(percent = count/sum(count),
           !! rlang::enquo(column) := case_when(change_order ~ 
        reorder({{column}}, -count, FUN=identity), TRUE ~ factor({{column}})) ) %>%
    ggplot(aes(x={{column}}, y=count)) +
      xlab(title)+
      geom_col(fill='#cccccc') +
      geom_text(aes(label = paste0(round(100 * percent, 1), "%")))



    }

-testing

my_bar_chart(mydata, Role, "EA5 Controller Role", TRUE)

enter image description here

Note that the reorder is creating some attributes to the concerned column which would be dropped within the case_when as it expects both the expressions to have the same structure output. Instead, we can use if/else here

my_bar_chart <- function(data, column, title, change_order=FALSE){
 out <-  data %>% 
    group_by({{column}}) %>% 
    summarize(count = n()) %>% 
    mutate(percent = count/sum(count),
           !! rlang::enquo(column) := if(change_order)
        reorder({{column}}, -count, FUN=identity) else {{column}} )

 print(str(out))
 ggplot(out, aes(x={{column}}, y=count)) +
   xlab(title)+
   geom_col(fill='#cccccc') +
   geom_text(aes(label = paste0(round(100 * percent, 1), "%")))








    }

-testing

Check the print output of the structure of dataset

my_bar_chart(mydata, Role, "EA5 Controller Role", TRUE) 
#Classes ‘tbl_df’, ‘tbl’ and 'data.frame':  2 obs. of  3 variables:
# $ Role   : Factor w/ 2 levels "Primary","Secondary": 1 2
#  ..- attr(*, "scores")= int [1:2(1d)] -3 -1
#  .. ..- attr(*, "dimnames")=List of 1
#  .. .. ..$ : chr  "Primary" "Secondary"
# $ count  : int  3 1
# $ percent: num  0.75 0.25
NULL

my_bar_chart(mydata, Role, "EA5 Controller Role", FALSE)
#Classes ‘tbl_df’, ‘tbl’ and 'data.frame':  2 obs. of  3 variables:
# $ Role   : chr  "Primary" "Secondary"
# $ count  : int  3 1
# $ percent: num  0.75 0.25

Upvotes: 1

Related Questions