3
$\begingroup$

Is there a complete list of all available models within fable package?

https://rdrr.io/cran/fabletools/man/model.html
model(
    snaive = SNAIVE(Turnover),
    ets = ETS(log(Turnover) ~ error("A") + trend("A") + season("A")),
    arima = ARIMA(Turnover)
  )

From the examples, only 3 models are available, which I find quite limited compared to the forecast package.

Improve this question
$\endgroup$

2 Answers 2

Reset to default
4
$\begingroup$

The reference page on the fable website contains an organised list of models: https://fable.tidyverts.org/reference/

Improve this answer
$\endgroup$
4
  • $\begingroup$ Hi Mitchelle, do you also have list of models which are supported by fable syntax? I'm aware of prophet fable, but hopefully want to discover others as well. I like the unified approach that fable is developing, though I think there're still some models which are not yet supported within fable or someone has developed it but not documented within fable reference page (prophet as an example) $\endgroup$
    – Afiq Johari
    Commented Jun 20, 2020 at 10:44
  • $\begingroup$ Currently the additional extension packages are fasster, fable.prophet, and fable.bsts. As this list grows, a centralised list of packages like exts.ggplot2.tidyverse.org would be useful, but this does not yet exist. $\endgroup$
    – Mitchell O'Hara-Wild
    Commented Jun 21, 2020 at 0:50
  • $\begingroup$ Thanks for the fable.bsts! Remember reading the request from Github issue, but didn't know it was implemented. $\endgroup$
    – Afiq Johari
    Commented Jun 21, 2020 at 8:39
  • 1
    $\begingroup$ Thanks to David Holt for that one: github.com/davidtedfordholt/fable.bsts $\endgroup$
    – Mitchell O'Hara-Wild
    Commented Jun 22, 2020 at 9:03
3
$\begingroup$

The most reliable way of understanding the capabilities of a package is, as always its CRAN page. In the specific case of the fable package, we find its reference manual and two different vignettes, one introduction and one vignette on forecasting with transformations.

The reference manual in particular looks helpful. For instance, I see no less than ten fitted.*() functions:

> library(dplyr)
> ls(getNamespace("fable"), all.names=TRUE) %>%
+ grep(pattern="fitted",value=TRUE)
[1] "fitted.AR"         "fitted.ARIMA"      "fitted.croston"   
[4] "fitted.ETS"        "fitted.model_mean" "fitted.NNETAR"    
[7] "fitted.RW"         "fitted.TSLM"       "fitted.VAR"

True, not all of them are very different (fitted.AR() vs. fitted.ARIMA()), but there are quite a few more than you mention (e.g., fitted.NNETAR(), fitted.fable_theta() or fitted.croston()).

Also note that fable is the successor package to forecast, so over time, it will likely accumulate more models.

Improve this answer
$\endgroup$
1
  • $\begingroup$ Thanks for the new trick on using the namespace. I'm aware of CRAN and the reference page but I doubt its accuracy. For example, prophet is not found either on the reference page nor in the fable reference page. Theta is also named differently, in reference page it's THETA but in the namespace it's fable_theta. $\endgroup$
    – Afiq Johari
    Commented Jun 20, 2020 at 10:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.