Skip to content

Validator#

Validators take an instance of a Learner, a Labeled dataset object, and a validation Metric and return a validation score that measures the generalization performance of the model using one of various cross validation techniques.

Note

There is no need to train the learner beforehand. The validator will automatically train the learner on subsets of the dataset created by the testing algorithm.

Test a Learner#

To train and test a Learner on a dataset and return the validation score:

public test(Learner $estimator, Labeled $dataset, Metric $metric) : float

use Rubix\ML\CrossValidation\KFold;
use Rubix\ML\CrossValidation\Metrics\Accuracy;

$validator = new KFold(10);

$score = $validator->test($estimator, $dataset, new Accuracy());

echo $score;
0.86

Last update: 2021-06-06