0

I have 600 series and I want to calculate the MASE for horizons 1, 3, 6, 8, 10 and 12. I've seen the work of Nikolaos Kourentzes where he calculates an ASE (no Mean) for each of three time horizons for thousands of series and then averages the ASE values across all series for each horizon to get a single MASE value. I thought the MASE was calculated only for each series, but now I'm confused.

Can I also calculate the ASE for each series at each horizon and then average across all series?

Angus
  • 239
  • You can calculate any measure of forecast accuracy that makes sense to you, but I would avoid changing the meanings of established terms such as MASE. That would introduce unnecessary confusion. – Richard Hardy May 01 '18 at 08:43
  • But my concern was average the Absolute Scaled Error (ASE) over all series at each horizon to get a single number rather than calculating a single MASE for each of 600 seres to get 600 MASE values.. Is that even valid? – Angus May 01 '18 at 11:16
  • It depends on what you want to estimate. The averaged ASE would be an estimator of its population counterpart. If you are interested in it, go ahead, it is your decision. – Richard Hardy May 01 '18 at 11:31

0 Answers0