Accuracy & evaluation

MAE

Also called mean absolute error

MAE is the average of |forecast − actual| in the original units. If every missed unit costs about the same, it is the everyday score that matches the cost.

No percentages, no weighting tricks. You miss by 12 units on average. Ops can feel that. Finance can multiply by price if they must.

MAE does not punish a single disaster more than RMSE does. If the rare spike is the whole risk, MAE will look calm while the warehouse burns. Know your loss.

Because it is scale-dependent, you cannot compare MAE across SKUs without grouping or scaling. That is what MASE and WAPE are for.