RandomSplit

One of the simplest methods from Cross-validation is implemented as RandomSpilt class. Samples are split to two groups: train group and test group. You can adjust number of samples in each group.

Constructor Parameters

• $dataset - object that implements Dataset interface
• $testSize - a fraction of test split (float, from 0 to 1, default: 0.3)
• $seed - seed for random generator (for tests)

$randomSplit = new RandomSplit($dataset, 0.2);

Samples and labels groups

To get samples or labels from test and train group you can use getters:

$dataset = new RandomSplit($dataset, 0.3, 1234);

// train group
$dataset->getTrainSamples();
$dataset->getTrainLabels();

// test group
$dataset->getTestSamples();
$dataset->getTestLabels();