Suggestion for new package covering missing stat functions

[msberends]

Would it be an idea to add another package with stat functions you would expect to be in stats but they are not?

Harmonic mean, geometric mean, mode, correlation of variance, standard errors, these kind of things? I could think of tens of obvious functions.

[bwiernik]

I'm not sure exactly what you mean by correlation of variance or standard errors. Could you elaborate?

[msberends]

As far as I know, there is no cv() function such as:

# coefficient of variation
cv <- function(..., na.rm = FALSE) {
  sd(c(...), na.rm = na.rm) / mean(..., na.rm = na.rm)
}

Another example, the coefficient of quartile variation (CQV, example in other package):

cqv <- function(x, na.rm = FALSE) {
  q1 <- quantile(x, 0.25)
  q3 <- quantile(x, 0.75)
  (q3 - q1) / (q3 + q1)
}

Wouldn’t the easystats be a great place for missing, obvious stat functions?

[msberends]

Ah, sorry, I now see I wrote ‘correlation’, where I meant ‘coefficient’. Excuse me!

[DominiqueMakowski]

I don't think there's enough of these for now to warrant their own separate package, but they could probably be incorporated within others, such as for instance bayestestR or parameters that have some functions to describe vectors

[bwiernik]

I think these could go in either parameters or datawizard

[IndrajeetPatil]

Yes, this would be a nice addition. I would vote for including them in datawizard.

[strengejacke]

CV (currently also in sjstats) is something that would fit into performance, since it somewhat close to r2() or rmse().

[strengejacke]

Maybe also some of the summary statistics from sjstats can be implemented in datawizard?

[msberends]

I don't think there's enough of these for now to warrant their own separate package

This could use some investigation. I’d like to think the opposite, but I’ll look around for blogs and question of people missing such functions in base R or ‘stats’.

[IndrajeetPatil]

Maybe also some of the summary statistics from sjstats can be implemented in datawizard?

Yes, a few of them can definitely be implemented.

Also, the weighted functions you have might also be helpful to be included in datawizard.

[bwiernik]

The describe distribution functions in general could be moved from parameters to datawizard (and re-exported).

[IndrajeetPatil]

Some of these functions have already been added in this repo, while there are open issues for the rest (#232, easystats/performance#433, etc.).

So I am going to close the current issue.