1
$\begingroup$

I am trying to relate data and results in NIST website with the formula defined in previous page from the same website. But I am missing something here:

  1. Does initial trend & season indices computation mean $b(1)$ and $I(1)$?

    If so, which one from the table of "initial seasonal indices" will be considered as $I(1)$?

  2. Let us say we are running this formula for t=1 and forecasting for t=4. How do we calculate $S(1)$ in this case? Won't $I(1-4)$ will be ZERO for $t=1$?

Any input would be appreciated.

Improve this question
$\endgroup$

1 Answer 1

Reset to default
1
$\begingroup$

The initial season indices (I(1) to I(4)) are computed in Step 3. As for the trend they take an average of the difference seen in the section "Initial values for the trend factor" Both can be found in the previous page. There are also multiple ways to get the initial value so results can be different depending on which method you use.

You can't Forecast from t=1 you need at least 2 at least periods or business cycles before you can determine the seasonal component.

Improve this answer
$\endgroup$
5
  • $\begingroup$ Thanks for the answer. I am forecasting for period 4, which means (F (t+m)) is 4, and m is 3 which leaves t=1. Isn't it? $\endgroup$
    – kosa
    Commented Feb 27, 2014 at 15:02
  • $\begingroup$ Just to add more info on why I am forecasting for period 4 is, if we observe first image in page2, triple started at 4, that means F(t+m) is 4, isn't it? $\endgroup$
    – kosa
    Commented Feb 27, 2014 at 16:20
  • $\begingroup$ @Nambari For this to work you have to start at a minimum of t=8 because you need at least 2 periods to set the season index or it will not work. $\endgroup$
    – ccsv
    Commented Mar 7, 2014 at 9:07
  • $\begingroup$ If we should start with t=8, how "triple" started at 4 in that image. This is the point I am stuck and going no where. Thanks for your time though. $\endgroup$
    – kosa
    Commented Mar 7, 2014 at 21:39
  • $\begingroup$ @Nambari You can not start at t=4 there is not enough data for that. $\endgroup$
    – ccsv
    Commented Mar 8, 2014 at 2:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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