K-Means Clustering Algorithm for PHP

A PHP implementation of the k-means clustering algorithm.

This algorithm groups data into clusters, based on the within-cluster sum of squares (intuitively, the squared Euclidean distance in the case of two-dimensional analysis). For example, a dataset containing the locations of many fish could be grouped into discrete schools of fish.

Limitations

• This library currently only supports two-dimensional clustering, but hopefully soon it will allow analysis on an arbitrary number of dimensions.
• The number of clusters to be used in the analysis must be manually configured.

Usage

Here is some example data:

$data = array(
	array(
		'name'	=> 'Foo',
		'x'		=> 3,
		'y'		=> 4
	),
	array(
		'name'	=> 'Bar',
		'x'		=> 5,
		'y'		=> 3
	),
	array(
		'name'	=> 'Baz',
		'x'		=> 12,
		'y'		=> 16
	),
)

We want to perform clustering on the x and y keys, and this data has two (very small) clusters. We configure the class like this:

$kmeans = new KMeans();
$kmeans
	->setData($data)
	->setXKey('x')
	->setYKey('y')
	->setClusterCount(2)
	->solve()
	

We retrieve the clusters like this:

$clusters = $kmeans->getClusters();

A cluster has data on its centroid:

foreach ($clusters as $cluster) {
	$x = $cluster->getX();
	$y = $cluster->getY();
}

And the points belonging to it:

$data = $cluster->getData();

Note that the original data arrays are preserved, so in our example, we could do this:

// print names of the objects in the first cluster
$names = array();
foreach ($clusters[0]->getData() as $ra) {
	$names[] = $ra['name'];
}
echo implode(', ', $names);

Which would output Foo, Bar.

License

This library is released under the MIT license. You may use it in personal or commercial projects. For more information, see the license file.