MASE
A scale-independent error metric widely used in time-series forecasting. It divides the absolute forecast error by the mean absolute error of a naive baseline (typically a random walk) computed on the training set, enabling cross-dataset comparisons.
Timeline (last 90 days)
On the GIFT-Eval benchmark, the best orchestrator reduces MASE by less than 6% relative to the best-performing single TSFM.
A MASE (mean absolute scaled error) below 1 indicates that the model outperforms the seasonal naive baseline, and it allows horizontal comparison and averaging across series of different magnitudes and frequencies.
The developer built a validation toolkit that includes forward rolling validation splits, block bootstrap and interval bootstrap procedures, permutation tests, and the WAPE and MASE evaluation metrics.
MASE divides the model's MAE by the naive forecast's MAE; a value less than 1 indicates it outperforms the naive model, and it is not affected by data scale, making it suitable for comparison across series.
MASE (Mean Absolute Scaled Error) is a dimensionless error metric commonly used in time series forecasting, dividing the absolute error by the mean absolute error of a naive baseline on the training set.
All Facts (5)
On the GIFT-Eval benchmark, the best orchestrator reduces MASE by less than 6% relative to the best-performing single TSFM.
50%UnverifiedA MASE (mean absolute scaled error) below 1 indicates that the model outperforms the seasonal naive baseline, and it allows horizontal comparison and averaging across series of different magnitudes and frequencies.
50%UnverifiedThe developer built a validation toolkit that includes forward rolling validation splits, block bootstrap and interval bootstrap procedures, permutation tests, and the WAPE and MASE evaluation metrics.
50%UnverifiedMASE divides the model's MAE by the naive forecast's MAE; a value less than 1 indicates it outperforms the naive model, and it is not affected by data scale, making it suitable for comparison across series.
50%UnverifiedMASE (Mean Absolute Scaled Error) is a dimensionless error metric commonly used in time series forecasting, dividing the absolute error by the mean absolute error of a naive baseline on the training set.
50%