Expected MASE

The Expected MASE (Mean Absolute Scaled Error) is the evaluation of the error made when using the predictive model to estimate the future values of the target, whatever the horizon.

The MASE is the MAE of the model (mean of the absolute differences between actual and forecasted values), divided by the MAE of a naive lag1 model (that is a model that would always use the last known value as prediction). The MASE provides a comparison between the model performance and the naive model performance.

The lower the MASE, the better the model performance. An Expected MASE greater than 1 indicates that the model performance is worse than that of the naive model.