Mitchell O'Hara-Wild

@earowang I don't think `difftime` can be used because it doesn't support irregular units such as months. Does this sound reasonable?

I have ideas for a new temporal vctrs object for handling temporal aggregation (https://github.com/tidyverts/fabletools/issues/59). This new temporal class will also support origin-less time offsets which ACF will use. It will...

Wait for https://github.com/mitchelloharawild/moment to provide origin-less time structures

This is an inherited from the {vctrs} package, the relevant issue for this is here: https://github.com/r-lib/vctrs/issues/1372 Passed via `…` ``` r library(tsibble) max(yearmonth('2017-01-01'), yearmonth('2018-04-01')) #> #> [1] "2017 Jan" ```...

This can be done either by suggesting `{tseries}` or re-implementing the method in feasts. For now it seems reasonable to use `tsfeatures::nonlinearity()` with `features()`.

Ideally this would give you a year-less unit, much like https://github.com/r-lib/clock/issues/61 as this would allow us to embed the "quarter" component to this. I don't know of an established data...

Yes, this is what I would like also. To make this possible, we need `features()` to provide the index variable to feature functions where necessary. In combination with potential improvements...

Each of the 'specials' in the model should be added to the formula. So instead of `X_13ARIMA_SEATS(Employed ~ transform(func = "none"), x11(mode = "add"))`, it should be `X_13ARIMA_SEATS(Employed ~ transform(func...

You're welcome. The backticks are tricky for this, because of course `function` has special parsing for creating functions in R. However I do think it is important to match the...

The decomposition methods in this package are less problematic as they can be (and are) used for calculating features. Additionally they have lighter package dependencies, or have dependencies listed as...