Function can't find column name in tibble - r

Question

I'm trying to create a function that uses dplyr syntax to manipulate data, but the function can't find the column names.
# example code below
library(dplyr)
# create sample data
ex.dat = data.frame(ex.IV = c(rep(1,50),
rep(2,50)),
ex.DV = c(rnorm(n = 50, mean = 100, sd = 15),
rnorm(n = 50, mean = 115, sd = 15)))
# create simple function that finds mean and sd from sample data
ex.func = function(data,predictor,predicted){
as.tibble(data) %>%
group_by(predictor) %>%
summarise(
M = mean(predicted),
SD = sd(predicted)
)
}
# run function with sample data
ex.func(data = ex.dat, predictor = ex.IV, predicted = ex.DV)
This produces the following error: "Error: Must group by variables found in .data. Column predictor is not found."
I don't understand why the function isn't assigning ex.IV to predictor.
Running the same code without involving a function, of course, has no issues, e.g.,
as.tibble(ex.dat) %>%
group_by(ex.IV) %>%
summarise(
M = mean(ex.DV),
SD = sd(ex.DV))
produces the intended result, so the issue must reside in the function formatting.
Workarounds like:
ex.func(data = ex.dat, predictor = ex.dat$ex.IV, predicted = ex.dat$ex.DV)
ex.func(data = ex.dat, predictor = data$ex.IV, predicted = data$ex.DV)
receive the same errors.
Clearly I'm not understanding some basic operations of function(). I'd appreciate some pointers.

Answer 1

We could make use of the curly-curly ({{}}) operator as the input argument are unquoted
ex.func <- function(data, predictor, predicted){
as.tibble(data) %>%
group_by({{predictor}}) %>%
summarise(
M = mean({{predicted}}),
SD = sd({{predicted}})
)
}
Now run as
ex.func(data = ex.dat, predictor = ex.IV, predicted = ex.DV)
If we need flexible option is the argument passed can be either quoted or unquoted, then we may need to convert to symbol with ensym and evaluate (!!
ex.func <- function(data, predictor, predicted){
predictor <- rlang::ensym(predictor)
predicted <- rlang::ensym(predicted)
as.tibble(data) %>%
group_by(!!predictor) %>%
summarise(
M = mean(!!predicted),
SD = sd(!!predicted)
)
}
Then, we can call either as
ex.func(data = ex.dat, predictor = ex.IV, predicted = ex.DV)
Or
ex.func(data = ex.dat, predictor = "ex.IV", predicted = "ex.DV")