Metrics

Validation metrics are for used evaluating the performance of an estimator. They output a score based on the predictions and the ground-truth found in the labels.

Scoring Predictions

To compute a validation score, pass in the predictions from an estimator along with the expected labels:

public score(array $predictions, array $labels) : float

use Rubix\ML\CrossValidation\Metrics\MeanAbsoluteError;

$predictions = $estimator->predict($dataset);

$metric = new MeanAbsoluteError();

$score = $metric->score($predictions, $dataset->labels());

echo $score;

-0.99846

Note

Regression metrics output the negative of their value to maintain the notion that cross validation scores should be maximized instead of minimized such as the case with loss functions.

Output Range

Output the minimum and maximum value the validation score can take in a 2-tuple:

public range() : Rubix\ML\Tuple{float, float}

[$min, $max] = $metric->range()->list();

echo "min: $min, max: $max";

min: -INF, max: 0

---

Last update: 2021-06-06