difficulty 1/5
MAE
The average size of the errors, ignoring their direction — the plain, honest answer to "how far off is this model typically?"
Intuition
Take the size of each miss, ignore whether it was over or under, and average them. That is MAE, and unlike RMSE it really is the average error — the number most people think they are being given when shown a regression score.
Where MSE squares the errors and so treats one miss of 10 as worse than ten misses of 1, MAE treats them as identical. Neither is more correct; they encode different beliefs about what a bad prediction costs.
Formal definition
How it works
Where it's used
Why & when
Worked example
Pitfalls
Check yourself
3 questions
The questions on this page come with access. They are graded on the server, so the answers stay there until you have given yours.
On the cheat sheet: Regression metrics and losses →
Sources
- scikit-learn — sklearn.metrics.mean_absolute_errordocs · retrieved 2026-08-30https://scikit-learn.org/stable/modules/generated/sklearn.metrics.mean_absolute_error.html
- Dive into Deep Learning — 3.1 Linear Regressionbookhttps://d2l.ai/chapter_linear-regression/linear-regression.html
- Deep Learning (2016)bookhttps://www.deeplearningbook.org/
- scikit-learn — sklearn.linear_model.HuberRegressordocs · retrieved 2026-10-06https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.HuberRegressor.html