|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* |
|
4
|
|
|
* (c) Ruben Dorado <[email protected]> |
|
5
|
|
|
* |
|
6
|
|
|
* For the full copyright and license information, please view the LICENSE |
|
7
|
|
|
* file that was distributed with this source code. |
|
8
|
|
|
*/ |
|
9
|
|
|
namespace SiteAnalyzer; |
|
10
|
|
|
|
|
11
|
|
|
use Exception; |
|
12
|
|
|
|
|
13
|
|
|
/** |
|
14
|
|
|
* class ML |
|
15
|
|
|
* |
|
16
|
|
|
* @package SiteAnalyzer |
|
17
|
|
|
* @author Ruben Dorado <[email protected]> |
|
18
|
|
|
* @copyright 2018 Ruben Dorado |
|
19
|
|
|
* @license http://www.opensource.org/licenses/MIT The MIT License |
|
20
|
|
|
*/ |
|
21
|
|
|
class ML |
|
22
|
|
|
{ |
|
23
|
|
|
|
|
24
|
|
|
/* |
|
25
|
|
|
* @param |
|
26
|
|
|
*/ |
|
27
|
|
|
function kmeans($data, $nclusters) |
|
|
|
|
|
|
28
|
|
|
{ |
|
29
|
|
|
$resp = []; |
|
30
|
|
|
$finished = false; |
|
31
|
|
|
$npoints = count($data); |
|
32
|
|
|
if (count($npoints) <= 0) throw new Exception("Not enough data. "); |
|
|
|
|
|
|
33
|
|
|
$ndimensions = count($npoints[0]); |
|
34
|
|
|
$centroids = initCentroids($nclusters, $ndimensions, function(){return rand(0,100)/100;}); |
|
|
|
|
|
|
35
|
|
|
while (!$finished) { |
|
36
|
|
|
|
|
37
|
|
|
// Assign each one of the points to one centroid |
|
38
|
|
|
$nresp = []; |
|
39
|
|
|
for ($j = 0; $j < $npoints; $j++) { |
|
40
|
|
|
$best = -1; |
|
41
|
|
|
$bdist = INF; |
|
42
|
|
|
for ($i = 0; $i < $nclusters; $i++) { |
|
43
|
|
|
$ndist = eclideanDistance($npoints[$j], $nclusters[$i]); |
|
|
|
|
|
|
44
|
|
|
if($bdist > $ndist) { |
|
45
|
|
|
$bdist = $ndist; |
|
46
|
|
|
$best = $i; |
|
47
|
|
|
} |
|
48
|
|
|
} |
|
49
|
|
|
$nresp[] = $best; |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
// Check change |
|
53
|
|
|
if(count($resp)!=0) { |
|
54
|
|
|
$finished = true; |
|
55
|
|
|
for ($j = 0; $j < $npoints; $j++) { |
|
56
|
|
|
if($resp[$j]!==$nresp[$j]){ |
|
57
|
|
|
$finished = false; |
|
58
|
|
|
break; |
|
59
|
|
|
} |
|
60
|
|
|
} |
|
61
|
|
|
$resp = $nresp; |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
/** Recalculate the centroids*/ |
|
65
|
|
|
$centroids = initCentroids($nclusters, $ndimensions, function(){return 0;}); |
|
66
|
|
|
$counts = array_fill(0, $nclusters, 0); |
|
67
|
|
|
for ($j = 0; $j < $npoints; $j++) { |
|
68
|
|
|
sumCentroid($centroids[$resp[$j]], $data[$j]); |
|
|
|
|
|
|
69
|
|
|
$counts[$resp[$j]]++; |
|
70
|
|
|
} |
|
71
|
|
|
normalizeCentroids($centroids, $counts); |
|
|
|
|
|
|
72
|
|
|
} |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
/* |
|
76
|
|
|
* @param |
|
77
|
|
|
*/ |
|
78
|
|
|
function initCentroids($nclusters, $ndimensions, $fvalue) |
|
|
|
|
|
|
79
|
|
|
{ |
|
80
|
|
|
$resp = []; |
|
81
|
|
|
for ($i = 0; $i < $nclusters; $i++) { |
|
82
|
|
|
$centroid = []; |
|
83
|
|
|
for ($d = 0; $d < $ndimensions; $d++) { |
|
84
|
|
|
$centroid[] = $fvalue(); |
|
85
|
|
|
} |
|
86
|
|
|
$resp[] = $centroid; |
|
87
|
|
|
} |
|
88
|
|
|
return $resp; |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
/* |
|
92
|
|
|
* @param |
|
93
|
|
|
*/ |
|
94
|
|
|
function eclideanDistance($p1, $p2) { |
|
|
|
|
|
|
95
|
|
|
$len = count($p1); |
|
96
|
|
|
$acum = 0; |
|
97
|
|
|
for($i=0; $i<$len; $i++) { |
|
98
|
|
|
$acum += ($p1[$i] - $p2[$i])**2; |
|
99
|
|
|
} |
|
100
|
|
|
return sqrt($acum); |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
|
|
} |
|
104
|
|
|
|
rdorado /
site-analyzer
Loading history...
Adding explicit visibility (
private,protected, orpublic) is generally recommend to communicate to other developers how, and from where this method is intended to be used.