Computer scienceData scienceMachine learningRegressionRegression metrics

MAE

The absolute value

Report a typo

We are summing the absolute values from the difference between the ground truth and the prediction. Which statements can be directly explained by taking the absolute values, instead of, say, squaring them?

Select one or more options from the list

MAE is a piece-wise linear function

The bias (how skewed the model is) can't be determined

The score is always non-negative

The fact that large and small errors don't contribute proportionally towards the score

___

Create a free account to access the full topic