Add missing select helpers

[etiennebacher]

Close #133

Still need to do demean() but I think it should be renamed data_demean() since it only accepts dataframes

[etiennebacher]

Still need to do demean() but I think it should be renamed data_demean() since it only accepts dataframes

@IndrajeetPatil @strengejacke do you agree on this?

[IndrajeetPatil]

Still need to do demean() but I think it should be renamed data_demean() since it only accepts dataframes

@IndrajeetPatil @strengejacke do you agree on this?

Sounds reasonable to me, but we will need to do this in a backward-compatible manner.

We can add deprecation message for now, and remove the older version in a future release.

[strengejacke]

Still need to do demean() but I think it should be renamed data_demean() since it only accepts dataframes

@IndrajeetPatil @strengejacke do you agree on this?

It's a decision between "accepts only data frames" and "belongs to the center and standardize family". You could argue therefore to keep this name.

[etiennebacher]

Whether we rename it or not, introducing select helpers in demean() would lead to breaking changes because we allow interaction between selected variables, which is not supported by .select_nse() at the moment (and maybe never will?):

library(datawizard)

dat <- data.frame(
  a = c(1, 2, 3, 4),
  x = c(4, 3, 3, 4),
  y = c(1, 2, 1, 2),
  ID = c(1, 2, 3, 1)
)
demean(dat, select = c("a", "x*y"), group = "ID")
#>   a_between x_y_between a_within x_y_within
#> 1       2.5           6     -1.5         -2
#> 2       2.0           6      0.0          0
#> 3       3.0           3      0.0          0
#> 4       2.5           6      1.5          2

[strengejacke]

I think we can just leave demean() as is. select-helpers are probably not that necessary here, because you usually have a limited set of variables in your model and even less that you want to demean (in practice). I think demeaning interaction terms (i.e. allow interactions in select) is here more important that the select-helper feature.

[etiennebacher]

Ok, demean() won't change and winsorize() (also in the list of #133) is still being worked on in other PRs so I don't modify it here.