juniorCost Function
What is Mean Absolute Error and when is it used?
Updated May 15, 2026
Short answer
MAE measures average absolute differences between predictions and actual values.
Deep explanation
MAE treats all errors equally by taking absolute differences, making it more robust to outliers than MSE.
Real-world example
Used in forecasting sales where occasional spikes should not dominate training.
Common mistakes
- Assuming MAE is always better than MSE.
Follow-up questions
- Is MAE differentiable?
- Why is MAE more robust?