[R] forecast method in stats package
|v@n @end|ng |rom @vetunkov@ru
Tue Aug 31 22:25:35 CEST 2021
I'm not sure where to post the suggestions for stats R package, so I've
decided to use the contact email from the package description.
There are several R packages that use "forecast()" method for purposes of
prediction (it is a bit different than "predict()"). The generic was
initially defined in "forecast" package, but now exists also in
"fabletools" and "greybox" (the two I know of, there might be others). And
as mentioned above, several packages implement the method for their
purposes (e.g. tsutils, nnfor, robets, fable, fasster, smooth etc), and
there are potential conflicts, when "forecast", "fabletools" and "greybox"
are used in the same script. So, given the popularity of the method,
wouldn't it make sense to implement the generic in "stats" package? The
default method could just call "predict.default()" with parameters provided
by the user. This would simplify the life of several package maintainers
and resolve the conflicts between packages. If needed, I could provide the
description of the method. Let me know what you think.
[[alternative HTML version deleted]]
More information about the R-help