I am trying to do a multivariate forecasting with ARIMAX. And I am not sure if ARIMAX can handle time series features as exogenous variables or it can only handle none time series features or both.


ARIMAX and regression with ARIMA errors (there is a difference) handle predictors that vary with time. It does not do anything with features that do not vary with time. (Any effects that do not vary with time will be absorbed into the intercept.)

  • 2
    $\begingroup$ so if I am doing a sales forecasting and I have some other variables (except from lagged history sales values) like promotion and day of the week. Can these be added as exogenous variables? or only another time series like feature (something like sales which is contentious) can only be added? I didn't get you properly. $\endgroup$ – Suleka_28 Dec 2 '18 at 14:05
  • 1
    $\begingroup$ Promotion and day of week are time series. The value of the predictor changes over time. Yes, you can include these as predictors. $\endgroup$ – Stephan Kolassa Dec 2 '18 at 14:35

Your Answer

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

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