seasonal dummies in ARIMA and the library forecast

[newsagent]

Using a monthly dataset from second edition of the book (link), suppose that

library(forecast)
fit1 <- Arima(insurance$TVadverts, order=c(0,0,2))

If we wanted to include seasonal dummies, how could we do this? Of course we have the option of SARIMA but I am interested in using seasonal dummies. I think that we should use xreg but I do not know how to incorporate these dummies in xreg.

[robjhyndman]

Use the seasonaldummy() function:

library(fpp2)
#> ── Attaching packages ──────────────────────────────────────────── fpp2 2.5.1 ──
#> ✔ ggplot2   4.0.3     ✔ fma       2.5  
#> ✔ forecast  9.0.2     ✔ expsmooth 2.3
#> 
fit1 <- Arima(insurance[, "TV.advert"],
  xreg = seasonaldummy(insurance[, "TV.advert"]),
  order = c(0, 0, 2)
)
forecast(fit1, xreg = seasonaldummy(insurance[, "TV.advert"], h = 12))
#>          Point Forecast    Lo 80     Hi 80    Lo 95     Hi 95
#> May 2005       9.689146 8.401893 10.976400 7.720462 11.657830
#> Jun 2005       8.267309 6.685478  9.849140 5.848107 10.686511
#> Jul 2005       7.999618 6.374112  9.625125 5.513621 10.485616
#> Aug 2005       7.605722 5.980215  9.231228 5.119724 10.091719
#> Sep 2005       7.693269 6.067762  9.318775 5.207271 10.179267
#> Oct 2005       7.311492 5.685986  8.936999 4.825495  9.797490
#> Nov 2005       7.883501 6.257994  9.509007 5.397503 10.369498
#> Dec 2005       7.251027 5.625520  8.876533 4.765029  9.737024
#> Jan 2006       8.493880 6.868374 10.119387 6.007882 10.979878
#> Feb 2006       9.841441 8.215935 11.466948 7.355443 12.327439
#> Mar 2006       8.295398 6.669892  9.920905 5.809401 10.781396
#> Apr 2006       7.505733 5.880226  9.131239 5.019735  9.991730

^{Created on 2026-06-03 with reprex v2.1.1}

[newsagent]

Thank you so much!!