Default values of to_numeric()

[strengejacke]

Regarding easystats/easystats#434, I was thinking what would be a good default for the dummy_factors argument in to_numeric()? Currently, the default is TRUE, splitting one variable into multiple ones.

Personally, I think this - as default - is a bit odd and I would expect a "numeric" variant of the factor, what dummy_factors = FALSE returns by default.

WDYT?

[etiennebacher]

I don't have a strong opinion on this (especially since I don't use this function), but since the next release is 0.13.0 and already contains a breaking change, it might be a good time to change this behavior too.

[etiennebacher]

Btw I lost track a bit of whether we had a discussion on this but is there any reason for having coerce_to_numeric() and to_numeric() instead of to_numeric() only?

[strengejacke]

Not sure about the current state, but we had:
#206 (comment)

I think the functions behave slightly different, and we needed both, at some point decided to make it (or keep it) exported. The major difference is that coerce_to_numeric() only returns numeric values when character values or factor levels were "numeric" before. Else, returns in the input as is.

[strengejacke]

See

datawizard::coerce_to_numeric(c("1", "2", "3"))
#> [1] 1 2 3
datawizard::to_numeric(c("1", "2", "3"))
#> [1] 1 2 3

datawizard::coerce_to_numeric(c("1", "2", "A"))
#> [1] "1" "2" "A"
datawizard::to_numeric(factor(c("1", "2", "A")))
#>   1 2 A
#> 1 1 0 0
#> 2 0 1 0
#> 3 0 0 1
datawizard::to_numeric(factor(c("1", "2", "A")), dummy_factors = FALSE)
#> [1] 1 2 3

^{Created on 2024-10-02 with reprex v2.1.1}

[strengejacke]

(in the above example you may see why dummy_factors = TRUE is a strange default)

[strengejacke]

I don't have a strong opinion on this (especially since I don't use this function), but since the next release is 0.13.0 and already contains a breaking change, it might be a good time to change this behavior too.

Let me know, I can make the changes and commit them to #546.

[etiennebacher]

Let's keep the distinction between to_numeric() and coerce_to_numeric() then, and move the default to dummy_factors = FALSE. Can you make this in a separate PR? I'll push other changes to #546 and I'd like to avoid conflicts there

Edit: my bad, didn't see #547