describe_distribution(): How to omit Skewness and Kurtosis?

[mvuorre]

I can easily turn off IQR with describe_distribution(..., iqr = FALSE). So I would expect to turn off the skewness and kurtosis variables with e.g. describe_distribution(..., skewness = FALSE), but apparently cannot. Is this, or could it be, possible?

[etiennebacher]

The output of describe_distribution() is a dataframe so you can easily drop columns, for instance with data_remove():

x <- datawizard::describe_distribution(1:10)
x
#> Mean |   SD |  IQR |         Range | Skewness | Kurtosis |  n | n_Missing
#> -------------------------------------------------------------------------
#> 5.50 | 3.03 | 5.50 | [1.00, 10.00] |        0 |    -1.20 | 10 |         0

datawizard::data_remove(x, c("Skewness", "Kurtosis"))
#> Mean |   SD |  IQR |         Range |  n | n_Missing
#> ---------------------------------------------------
#> 5.50 | 3.03 | 5.50 | [1.00, 10.00] | 10 |         0

[strengejacke]

We probably should have made a different approach, using something like indices = c("iqr", "range", ...) or similar, because now we have an argument for almost each column, which makes the list of arguments quite long...

[mvuorre]

Ok thanks! Fwiw using options() to control the output might be nice to reduce repetitions, e.g.

options(datawizard.describe_skewness = FALSE

And then these could be used in the function(s).

[mattansb]

We probably should have made a different approach, using something like indices = c("iqr", "range", ...)

Should we make a little breaking change that will make this function also easily expandable?

Fwiw using options() to [...]

I think we hardly use options() and getOptions() across easystats for some reason?

[strengejacke]

We use options in some packages and have some documented options there:
https://easystats.github.io/parameters/reference/parameters-options.html
https://easystats.github.io/modelbased/reference/modelbased-options.html

There are maybe more options, not yet documented as "overview".

Should we make a little breaking change that will make this function also easily expandable?

I'd say yes. And maybe also having options, too, which is quite convenient.

[strengejacke]

See #685 and related PR #689, maybe we can have a similar approach.