When determining the baseline performance using a persistent or naive model I understand it to be using the value of the previous time step as the prediction for the next time step and then determining the accuracy.

My question is how would one do this when determining the baseline for a 7-day forecast? Would you calculate the accuracy using the previous time step (t-1) to predict only the next time step (t), or would you use the previous time step (t-1) to predict all 7 future time steps (t,t+1,t+2 etc).

  • $\begingroup$ Does your data exhibit a strong 7-day seasonal cycle ? $\endgroup$
    – AlainD
    Aug 18 '18 at 8:10
  • $\begingroup$ @AlainD not that I'm aware of $\endgroup$ Aug 20 '18 at 13:13
  • $\begingroup$ he it does not really matter. The baseline is just a very rough forecast whose only quality is that is it easy to compute. Take the easiest to compute. If in doubt, compute with both methods on a sample and keep the best one and don't think about it any more. $\endgroup$
    – AlainD
    Aug 21 '18 at 7:25

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.