blog/2022-11-07-lets-get-applying/

[giscus[bot]]

blog/2022-11-07-lets-get-applying/

https://drmowinckels.io/blog/2022-11-07-lets-get-applying/

[AlexWeinreb]

Nice post! tapply() also works quite well for this example:

tapply(mtcars$mpg, mtcars$cyl, \(x) x - mean(x))

[AlexWeinreb]

Yet another possibility that just came to my mind, this pattern is sometimes called split-apply-combine, and can also be done with:

split(mtcars$mpg, f = mtcars$cyl) |>
  lapply(\(x) x - mean(x)) |>
  unsplit(mtcars$cyl)

which, unlike my previous tapply() approach, does keep the order of the elements (of course your mapply() also did).

[drmowinckels]

I've never really got the hang of tapply so thanks for the example!
The second approach is also very interesting! I personally find mappy easier to read, and also easier to scale up to more paramenters, but its still pretty neat!

[psolymos]

Nice post, thanks for the summary! Some base R magic is always appreciated. If you need a progress bar and quickly accessing multiple cores, have a look at {pbapply} [apply(...) -> pbapply(..., cl=3)].

[drmowinckels]

I definitely have a post on progress-bars and on multi-core apply's in the making ;)
They are both very convenient when there are many items to apply through, or the function is computationally taxing.