Evaluating Regression Models

MAE is the average absolute difference between the predicted values and the actual values. It is a more robust metric than MSE, as it is not as sensitive to outliers. However, it is not as differentiable as MSE, which can make it more difficult to optimize.

• not very sensitive to outliers, there useful for datasets with many Outlier!
• Slow due to more complex computations (Erklärung p. 192 Size 22%)

Kaggle:

• LANL Earthquake Prediction
• How Much Did It Rain? II