jacobemerick/kmeans

k-means clustering implemented in PHP

Maintainers

Details

github.com/jacobemerick/kmeans

Source

Issues

Installs: 5 983

Dependents: 0

Suggesters: 0

Security: 0

Stars: 12

Watchers: 2

Forks: 4

Open Issues: 3

1.0.0 2014-07-31 00:59 UTC

Requires

  • php: >=5.3

MIT 74808b7b65546e96449b0c8cca66b42236c4a6d8

  • Jacob Emerick

statisticsclusteringk-meansdata analysis

This package is not auto-updated.

Last update: 2024-11-19 07:30:53 UTC

README

This handly little class will calculate the k-means for a set of observations using PHP. k-means is a cool way to cluster data into groups based on relation - like clustering geographical data (using lat/lng) into a digestible summary. It is useful for detecting patterns in large data sets.

Usage

Let's say that you wanted to cluster a data set. The data must be in a multi-dimensional array, each value a numeric, though the size of each row has no constraint (n-dimensions ftw).

$array = [
    [1, 1, 3],
    [3, 7, 6],
    [5, 8, 3],
    [1, 2, 1],
    [9, 10, 8],
    [4, 4, 4],
];

By observation you may suspect that this data can be clustered into 3 separate sets. To test, run the class.

$kmeans = new Jacobemerick\KMeans\Kmeans($array);
$kmeans->cluster(3); // cluster into three sets

$clustered_data = $kmeans->getClusteredData();
// $clustered_data = [
//     [[1, 1, 3], [1, 2, 1]],
//     [[3, 5, 6], [5, 4, 3], [4, 4, 4]],
//     [[9, 10, 8]],
// ];

$centroids = $kmeans->getCentroids();
// $centroids = [
//     [1, 1.5, 2],
//     [4, 4.33, 4.33],
//     [9, 10, 8],
// ];

Note: larger data sets will be more consistent - if you run this example multiple times your results may vary.

Installation

Through composer:

$ composer require jacobemerick/kmeans:~1.0