You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi,
it would be beneficial to expand the losses with the two official metrics used in the M5 Uncertainty competition, namely the Scaled Quantile Loss and the Scaled Multi-Quantile Loss.
From this reference https://www.sciencedirect.com/science/article/pii/S0169207021001722
the scaling factor is simply the in-sample, one-step-ahead mean absolute error of the Naive method. However, a more general version could be implemented using the Seasonal Naive method as is the case for MASE and RMSSE.
I can try to make a PR with the basis on this if you like.
Thanks!
Use case
No response
The text was updated successfully, but these errors were encountered:
Description
Hi,
it would be beneficial to expand the losses with the two official metrics used in the M5 Uncertainty competition, namely the Scaled Quantile Loss and the Scaled Multi-Quantile Loss.
From this reference https://www.sciencedirect.com/science/article/pii/S0169207021001722
the scaling factor is simply the in-sample, one-step-ahead mean absolute error of the Naive method. However, a more general version could be implemented using the Seasonal Naive method as is the case for MASE and RMSSE.
I can try to make a PR with the basis on this if you like.
Thanks!
Use case
No response
The text was updated successfully, but these errors were encountered: