unnormalize() with grouped data

[etiennebacher]

Close #375.

library(datawizard)

x <- iris |> 
  data_group(Species) |>
  normalize(select = Sepal.Length) |> 
  unnormalize(select = Sepal.Length) |> 
  data_ungroup()

identical(x, iris)
#> [1] TRUE

x <- iris |> 
  data_group(Species) |>
  standardize(select = Sepal.Length) |> 
  unstandardize(select = Sepal.Length) |> 
  data_ungroup()

identical(x, iris)
#> [1] TRUE

The idea is to store the dw_transformer attributes along the groups attributes so that we can then recover the dw_transformer attributes for each group separately. We then need to pass the dw_transformer attributes to normalize.data.frame() and then to normalize.numeric().

Same thing for unstandardize().

Note: there are probably ways to improve this in terms of performance

[etiennebacher]

@strengejacke if you're ok with this code then I can expand it to unstandardize()

[strengejacke]

I'm not sure, but I think @mattansb started with one of the un*() functions? Maybe you can better review this PR?

[mattansb]

Daniel keeps thinking that @DominiqueMakowski 's code is mine 😅

[etiennebacher]

The code is still incomplete for unstandardize(), it doesn't work with the example in #375

[strengejacke]

Daniel keeps thinking that @DominiqueMakowski 's code is mine 😅

But see effectsize news, version 0.5.0 unstandardize() was moved to datawizard, and I think you made one of the first implementations of those un*() functions (I think I then tried to adopt this for unnormalize()).

[mattansb]

I'm like 90% certain it was Dom, and then I maintained the code.

[DominiqueMakowski]

What Daniel says becomes true tho 🤷

[etiennebacher]

@strengejacke @DominiqueMakowski or @mattansb I don't know who's the best placed to review but can one of you do it? 😄

[etiennebacher]

Last thing, I think we should harmonize the behavior of unstandardize() and unnormalize() when they don't have the necessary info. Currently, unstandardize() fails but unnormalize() only gives a warning:

library(datawizard)

unstandardize(mtcars, "mpg")
#> Error: You must provide the arguments `center`, `scale` or `reference`.

unnormalize(head(mtcars), "mpg")
#> Warning: Can't unnormalize variable. Information about range and/or minimum value
#>   is missing.
#>                    mpg cyl disp  hp drat    wt  qsec vs am gear carb
#> Mazda RX4         21.0   6  160 110 3.90 2.620 16.46  0  1    4    4
#> Mazda RX4 Wag     21.0   6  160 110 3.90 2.875 17.02  0  1    4    4
#> Datsun 710        22.8   4  108  93 3.85 2.320 18.61  1  1    4    1
#> Hornet 4 Drive    21.4   6  258 110 3.08 3.215 19.44  1  0    3    1
#> Hornet Sportabout 18.7   8  360 175 3.15 3.440 17.02  0  0    3    2
#> Valiant           18.1   6  225 105 2.76 3.460 20.22  1  0    3    1

IMO both should fail. @strengejacke do you agree?

[strengejacke]

yes.

[etiennebacher]

Failures due to easystats/insight#804