R dplyr Quosure logic when function input is a variable

Question

I'm trying to get a custom group_by function working using quosure. It works fine when the input into the function is the name of a field. However, if I want to create a variable that contains the name of the field to insert into the function, I get stuck because the field is quoted. Anyone know how to get around this. Please see below

 groupedMean<-function(field){  
    expr<-enquo(field)

   mtcars%>%
       group_by(!!expr)%>%
       summarise(mean(mpg))  
 }

 #Works
 groupedMean(cyl)

 #Doesn't work
 groupFields<-c("cyl","gear")

 for(var in groupFields){
    print(groupedMean(eval(var)))
 }

akrun akrun · Accepted Answer · 2017-12-20T16:48:30

One option is to convert it using sym from rlang

for(var in groupFields){
     var <- rlang::sym(var)
     print(groupedMean(!!var))
}
# A tibble: 3 x 2
#    cyl `mean(mpg)`
#  <dbl>       <dbl>
#1     4    26.66364
#2     6    19.74286
#3     8    15.10000
# A tibble: 3 x 2
#   gear `mean(mpg)`
#  <dbl>       <dbl>
#1     3    16.10667
#2     4    24.53333
#3     5    21.38000

Instead of printing, for later use, the output can be stored in a list

lst <- setNames(vector('list', length(groupFields)), groupFields)
for(var in groupFields){
   var <- rlang::sym(var)
   lst[[var]] <- groupedMean(!!var)
}
lst

R dplyr Quosure logic when function input is a variable

1 Answers