Mean Absolute Error#

A scale-dependent metric that measures the average absolute error between a set of predictions and their ground truth labels. MAE has the same units of measurement as the labels being estimated.

Note: In order to maintain the convention of maximizing validation scores, this metric outputs the negative of the original score.

Estimator Compatibility: Regressor

Output Range: -∞ to 0


use Rubix\ML\CrossValidation\Metrics\MeanAbsoluteError;

$metric = new MeanAbsoluteError();