I am building a weekly forecasting model using dynamic harmonic regression model with Arima errors and Fourier terms. My understanding is that Fourier terms take care of seasonality of any length automatically. However, to account for the impact of holidays we need to define dummy variables manually, is that correct?

  • $\begingroup$ Yes, the model has no way of knowing what is a holiday. $\endgroup$ – user2974951 Oct 2 '19 at 8:58
  • $\begingroup$ But is it not the whole point of Fourier terms to capture the seasonality automatically. If Fourier terms captures more than one seasonality (if it exists), then why bother doing the dummy variables, in addition, which also aims to capture the seasonality due to the holiday impact? $\endgroup$ – Vusal Oct 3 '19 at 20:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.