Unquote character in `.vars` argument in `fabletools::autoplot_tbl_ts()`

Question

If I want to use a character vector in dplyr, I can do this:

library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
  library(rlang)
  library(tibble)
  
  my_var <- "mpg"
  my_data <- as_tibble(mtcars)
  
  my_data |>
    select(!!my_var)

^{Created on 2023-08-14 with reprex v2.0.2}

Likewise, I want to do the same in fabletools::autoplot.tbl_ts():

library(testthat)
  set.seed(12345)
  df <- tsibble::tsibble(
    qtr = tsibble::yearquarter("2010 Q1") + 0:59,
    y = rnorm(60),
    index = qtr
  )
  dv_name <- "y"
  
  actual <- fabletools::autoplot(df, !!dv_name)
  expected <- fabletools::autoplot(df, y)
  expect_equal(actual, expected)
#> Error: `actual` not equal to `expected`

^{Created on 2023-08-14 with reprex v2.0.2}

Seems like the bang-bang !! is not the right choice here. What should I do instead?

score 1 · Accepted Answer · answered Aug 14 '23 at 15:14

The problem is that dv_name is a string, not a symbol. Use

fabletools::autoplot(df, .data[[dv_name]])

That will look up the string column name in the data. The .data object is a special pronoun provided by rlang. The fact that this works with select() is kind of specific to that function. You can use both

select(mtcars, mpg)
select(mtcars, "mpg")

But using a string for a column name doesn't work the same in most other functions.

you could also translate the string into a symbol and then use !!. For example

fabletools::autoplot(df, !!rlang::ensym(dv_name))

Unquote character in `.vars` argument in `fabletools::autoplot_tbl_ts()`

1 Answers1