php-ml/tests/Phpml/CrossValidation/RandomSplitTest.php

95 lines
3.4 KiB
PHP
Raw Normal View History

2016-04-07 20:12:36 +00:00
<?php
2016-04-07 20:35:49 +00:00
declare (strict_types = 1);
2016-04-07 20:12:36 +00:00
namespace tests\Phpml\CrossValidation;
use Phpml\CrossValidation\RandomSplit;
use Phpml\Dataset\ArrayDataset;
class RandomSplitTest extends \PHPUnit_Framework_TestCase
{
/**
* @expectedException \Phpml\Exception\InvalidArgumentException
*/
public function testThrowExceptionOnToSmallTestSize()
{
new RandomSplit(new ArrayDataset([], []), 0);
}
/**
* @expectedException \Phpml\Exception\InvalidArgumentException
*/
public function testThrowExceptionOnToBigTestSize()
{
new RandomSplit(new ArrayDataset([], []), 1);
}
public function testDatasetRandomSplitWithoutSeed()
{
$dataset = new ArrayDataset(
$samples = [[1], [2], [3], [4]],
$labels = ['a', 'a', 'b', 'b']
);
2016-04-08 22:36:48 +00:00
$randomSplit = new RandomSplit($dataset, 0.5);
2016-04-07 20:12:36 +00:00
2016-04-08 22:36:48 +00:00
$this->assertEquals(2, count($randomSplit->getTestSamples()));
$this->assertEquals(2, count($randomSplit->getTrainSamples()));
2016-04-07 20:12:36 +00:00
$randomSplit2 = new RandomSplit($dataset, 0.25);
$this->assertEquals(1, count($randomSplit2->getTestSamples()));
$this->assertEquals(3, count($randomSplit2->getTrainSamples()));
}
public function testDatasetRandomSplitWithSameSeed()
{
$dataset = new ArrayDataset(
$samples = [[1], [2], [3], [4], [5], [6], [7], [8]],
$labels = ['a', 'a', 'a', 'a', 'b', 'b', 'b', 'b']
);
$seed = 123;
$randomSplit1 = new RandomSplit($dataset, 0.5, $seed);
$randomSplit2 = new RandomSplit($dataset, 0.5, $seed);
$this->assertEquals($randomSplit1->getTestLabels(), $randomSplit2->getTestLabels());
$this->assertEquals($randomSplit1->getTestSamples(), $randomSplit2->getTestSamples());
$this->assertEquals($randomSplit1->getTrainLabels(), $randomSplit2->getTrainLabels());
$this->assertEquals($randomSplit1->getTrainSamples(), $randomSplit2->getTrainSamples());
}
public function testDatasetRandomSplitWithDifferentSeed()
{
$dataset = new ArrayDataset(
$samples = [[1], [2], [3], [4], [5], [6], [7], [8]],
$labels = ['a', 'a', 'a', 'a', 'b', 'b', 'b', 'b']
);
$randomSplit1 = new RandomSplit($dataset, 0.5, 4321);
$randomSplit2 = new RandomSplit($dataset, 0.5, 1234);
$this->assertNotEquals($randomSplit1->getTestLabels(), $randomSplit2->getTestLabels());
$this->assertNotEquals($randomSplit1->getTestSamples(), $randomSplit2->getTestSamples());
$this->assertNotEquals($randomSplit1->getTrainLabels(), $randomSplit2->getTrainLabels());
$this->assertNotEquals($randomSplit1->getTrainSamples(), $randomSplit2->getTrainSamples());
}
public function testRandomSplitCorrectSampleAndLabelPosition()
{
$dataset = new ArrayDataset(
$samples = [[1], [2], [3], [4]],
$labels = [1, 2, 3, 4]
);
$randomSplit = new RandomSplit($dataset, 0.5);
$this->assertEquals($randomSplit->getTestSamples()[0][0], $randomSplit->getTestLabels()[0]);
$this->assertEquals($randomSplit->getTestSamples()[1][0], $randomSplit->getTestLabels()[1]);
$this->assertEquals($randomSplit->getTrainSamples()[0][0], $randomSplit->getTrainLabels()[0]);
$this->assertEquals($randomSplit->getTrainSamples()[1][0], $randomSplit->getTrainLabels()[1]);
}
2016-04-07 20:35:49 +00:00
}