I'm dealing with sales forecasts (time-series with few data points) for a company on a department level (one company consists out of multiple departments). My approach is it to build a time-series model for each department in order to forecast sales for each department. My problem is, I also want to know the overall sales of the company. Beside the solution to just train a time-series model on the aggragated overall company sales data, I wonder if I also can just add up the predictions from each department in order to create an overall company sales forecast. If I can do so, what actually happens with the confidence intervals? Does the boundries of the confidence intervals for each department add up?

  • $\begingroup$ It is not that obvious about the "confidence intervals". First of all, do you mean prediction intervals? Next, of course, you can build department specific models and add them up. In this way your additive model will contain possibly different autoregressive and seasonal structures. It can be more accurate than taking the total sales. $\endgroup$ – Alexey says Reinstate Monica Oct 16 at 13:33

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.