1
$\begingroup$

I have monthly sales data for about 7 years and i need to forecast sales figures for the next 10-15 years (yearly basis, not monthly). Also i need to use ARIMA for this.

How to approach this task properly? Should i build a model for the monthly sales figures, forecast all months of needed 15 years and than aggregate it to the level of years? Or is it better to convert my monthly time series to yearly and then make a forecast (however it will be only 7 data points for the needed forecast of 15)? Or maybe some combination?

Improve this question
$\endgroup$
3
  • 3
    $\begingroup$ If this is just an exercise, you can try playing with ARIMA. But if this is a real world problem, I doubt ARIMA would do any good in such long horizons. You should better check some forecasts for long-term industry trends and think how your particular product should fare relatively to those. $\endgroup$
    – Richard Hardy
    Commented Nov 21, 2016 at 9:38
  • $\begingroup$ @RichardHardy, yes, it is an exercise, so i need some ARIMA model or a combination. $\endgroup$
    – Koncopd
    Commented Nov 21, 2016 at 10:52
  • 2
    $\begingroup$ Do only you have 7 observations? If so, then there is essentially no chance of uncovering the dynamics using an ARMA model reliably. If you don't know the dynamics then a forecast will not be possible to provide even in the short-run (say 1 year) let alone for a horizon almost twice as large as your sample size. $\endgroup$
    – Math-fun
    Commented Oct 24, 2023 at 7:00

1 Answer 1

Reset to default
-1
$\begingroup$

"All models are wrong, some are useful." George Box said this and he created ARIMA models. ARIMA will work just fine. ARIMA models are a superset of all other models(exponential smoothing, etc.) except H-W Multiplicative. If you have a causal variable you will enhance your model and forecast. If you think the sales are tried to population or some other variable then by all means use it. If there is a life cycle to your product where it will become obsolete in 3 years then you can guide the forecast down by creating a dummy variable with a 1 in the history and right before the drop and then using .9, .8, .7 etc for example to guide the forecast down. You can also use analogues as causals to do something similar.

Don't sum your data to annual. Just sum the monthly forecasts to yearly when you are done.

Improve this answer
$\endgroup$
4
  • $\begingroup$ downvote? because? $\endgroup$
    – Tom Reilly
    Commented Nov 21, 2016 at 14:17
  • $\begingroup$ i didn't downvote your answer. $\endgroup$
    – Koncopd
    Commented Nov 21, 2016 at 19:53
  • $\begingroup$ ok, but somebody did :) $\endgroup$
    – Tom Reilly
    Commented Nov 21, 2016 at 20:16
  • $\begingroup$ I didn't downvote, but is ARIMA really a superset of exponential smoothing? The way that ARIMA incorporates trend through differencing seems less flexible and less sophisticated than an exponential smoothing trend term. $\endgroup$
    – Arthur
    Commented Feb 8, 2022 at 14:05

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.