Writing Functions With a “data” Argument

Question

I want to write a function that can take columns within a data frame or column names and the data frame they come from as arguments.

df <- data.frame(x = c(1:         


        

Answer 1

To sort of piggy-back off of Cettt - something like this may be what you're looking for:

df <- data.frame(x = c(1:5), y = c(6:10), z = LETTERS[1:5])

my_fxn <- function (aaa, bbb, ccc, data) {
  if (!missing(data)) {
    aaa = as.numeric(data[[aaa]])
    bbb = as.numeric(data[[bbb]])
    ccc = as.character(data[[ccc]])
  }
  print(aaa[1])
}

my_fxn("x", "y", "z", df)
#> [1] 1

---

With the use of enquo() from library(dplyr), we no longer need to enter characters as the function variables:

library(dplyr)

my_fxn <- function (aaa, bbb, ccc, data) {
  aaa <- enquo(aaa)
  bbb <- enquo(bbb)
  ccc <- enquo(ccc)

  if (!missing(data)) {
    aaa = as.numeric(pull(data, !!aaa))
    bbb = as.numeric(pull(data, !!bbb))
    ccc = as.character(pull(data, !!ccc))
  }
  print(aaa[1])
}

my_fxn(x, y, z, df)
#> [1] 1

More info about function building with enquo() and !! can be found here: https://dplyr.tidyverse.org/articles/programming.html#programming-recipes

---

Finally, a base R solution using deparse() and substitute():

my_fxn <- function (aaa, bbb, ccc, data) {
  aaa <- deparse(substitute(aaa))
  bbb <- deparse(substitute(bbb))
  ccc <- deparse(substitute(ccc))

  if (!missing(data)) {
    aaa = as.numeric(data[[aaa]])
    bbb = as.numeric(data[[bbb]])
    ccc = as.character(data[[ccc]])
  }
  print(aaa[1])
}

my_fxn(x, y, z, df)
#> [1] 1

Answer 2

The problem is that when calling my_fxn(x, y, z, df) the object x is not defined. Hence df$x does not return column x but NA.

Consider this small example:

df <- data.frame(x = 1:3, y = 4:6)
x <- "y"
df$x # returns column x
[1] 1 2 3
df[,x] #returns column y since the value which is stored in x is "y"
[1] 4 5 6

To circumvent your problem you can use data[, aaa] instead of data$aaa. Yet another alternative would be to use the dplyr package where you can use select(data, aaa).