mae

getml.pipeline.metrics.mae = 'mae'

Mean Absolute Error - measure of distance between two numerical targets.

Used for regression problems.

\[MAE = \frac{\sum_{i=1}^n | \mathbf{y}_i - \mathbf{\hat{y}}_i |}{n},\]

where \(\mathbf{y}_i\) and \(\mathbf{\hat{y}}_i\) are the target values or prediction respectively for a particular data sample \(i\) (both multidimensional in case of using multiple targets) while \(n\) is the number of samples we consider during the scoring.