tsorder model for the fable/tidyverts framework

A fabletools model definition: selects the ARIMA orders with a tsorder selector (via fit_arima()) and fits stats::arima(), so tsorder plugs directly into fabletools::model(), forecast(), accuracy() and friends alongside fable::ARIMA().

Usage

TSORDER(formula, method = "hr", ...)

Arguments

formula

Model specification (the response; exogenous regressors are not supported).

method

As in fit_arima(): an order_select() method or "hr" (default) for Hannan-Rissanen, which selects p and q jointly.

...

Passed on to fit_arima() (e.g. d, selector options).

Value

A model definition usable inside fabletools::model().

Details

Requires the suggested packages fabletools, tsibble and distributional.

Examples

if (FALSE) { # \dontrun{
library(fabletools)
as_tsibble(USAccDeaths) |>
  model(tsorder = TSORDER(value)) |>
  forecast(h = 12)
} # }