|
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
|
|
|
* class CategoricalDataset |
|
12
|
|
|
* |
|
13
|
|
|
* @package SiteAnalyzer |
|
14
|
|
|
* @author Ruben Dorado <[email protected]> |
|
15
|
|
|
* @copyright 2018 Ruben Dorado |
|
16
|
|
|
* @license http://www.opensource.org/licenses/MIT The MIT License |
|
17
|
|
|
*/ |
|
18
|
|
|
class CategoricalDataset |
|
19
|
|
|
{ |
|
20
|
|
|
|
|
21
|
|
|
/** |
|
22
|
|
|
* @var array |
|
23
|
|
|
*/ |
|
24
|
|
|
protected $data; |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* @var array |
|
28
|
|
|
*/ |
|
29
|
|
|
protected $sortedEncodedFeatures; |
|
30
|
|
|
|
|
31
|
|
|
/** |
|
32
|
|
|
* @var array |
|
33
|
|
|
*/ |
|
34
|
|
|
protected $encodedValues; |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* @var array |
|
38
|
|
|
*/ |
|
39
|
|
|
protected $featEncode; |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* @var array |
|
43
|
|
|
*/ |
|
44
|
|
|
protected $featIndexMap; |
|
45
|
|
|
|
|
46
|
|
|
/* |
|
47
|
|
|
* @param |
|
48
|
|
|
*/ |
|
49
|
|
|
public function __construct($data) |
|
50
|
|
|
{ |
|
51
|
|
|
$this->data = $data; |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
/* |
|
55
|
|
|
* @param |
|
56
|
|
|
*/ |
|
57
|
|
|
public function setEncodedFeatures($array) |
|
58
|
|
|
{ |
|
59
|
|
|
sort($array); |
|
60
|
|
|
$this->encodedValues = []; |
|
61
|
|
|
$this->sortedEncodedFeatures = $array; |
|
62
|
|
|
foreach($this->sortedEncodedFeatures as $col){ |
|
63
|
|
|
$vals = $this->getUniqueValues($col); |
|
64
|
|
|
$this->encodedValues[$col] = $vals; |
|
65
|
|
|
$this->featIndexMap[$col] = count($vals); |
|
66
|
|
|
$this->featEncode[$col] = $this->encodeFeature($vals); |
|
67
|
|
|
//$this->featDecode[$col] = function($val, $arr){ return $this->getDecodedFeature($val, $arr, ); } |
|
68
|
|
|
//$this->newEncodedSize += count($vals)-1; |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
/*for ($i=0;$i<$this->newEncodedSize:$i++) { |
|
72
|
|
|
|
|
73
|
|
|
}*/ |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
/* |
|
77
|
|
|
* @param |
|
78
|
|
|
*/ |
|
79
|
|
|
private function getUniqueValues($col) |
|
80
|
|
|
{ |
|
81
|
|
|
$tmp = Matrix::getColumn($this->data, $col); |
|
82
|
|
|
$n = count($tmp); |
|
83
|
|
|
$resp = []; |
|
84
|
|
|
for ($i=0; $i<$n; $i++) { |
|
85
|
|
|
if (!in_array($tmp[$i], $resp)) { |
|
86
|
|
|
$resp[] = $tmp[$i]; |
|
87
|
|
|
} |
|
88
|
|
|
} |
|
89
|
|
|
$resp = array_unique($resp); |
|
90
|
|
|
return $resp; |
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
|
|
/* |
|
94
|
|
|
* @param |
|
95
|
|
|
*/ |
|
96
|
|
|
private function encodeFeature($array) |
|
97
|
|
|
{ |
|
98
|
|
|
$size = count($array); |
|
99
|
|
|
$resp = []; |
|
100
|
|
|
for ($i=0;$i<$size;$i++) { |
|
101
|
|
|
$tmp = array_fill(0, $size, 0); |
|
102
|
|
|
$tmp[$i] = 1; |
|
103
|
|
|
$resp[$array[$i]] = $tmp; |
|
104
|
|
|
} |
|
105
|
|
|
return $resp; |
|
106
|
|
|
} |
|
107
|
|
|
|
|
108
|
|
|
/* |
|
109
|
|
|
* @param |
|
110
|
|
|
*/ |
|
111
|
|
|
public function encode(){ |
|
112
|
|
|
$transformer = []; |
|
113
|
|
|
$n = count($this->data); |
|
114
|
|
|
$ndim = count($this->data[0]); |
|
115
|
|
|
for ($j=0; $j<$ndim; $j++) { |
|
116
|
|
|
$transformer[] = function($val){ return [$val]; }; |
|
117
|
|
|
} |
|
118
|
|
|
foreach($this->sortedEncodedFeatures as $col) { |
|
119
|
|
|
$transformer[$col] = function ($val) use ($col) { return $this->featEncode[$col][$val]; }; |
|
120
|
|
|
} |
|
121
|
|
|
|
|
122
|
|
|
$ndata = []; |
|
123
|
|
|
for ($i=0; $i<$n; $i++) { |
|
124
|
|
|
$npoint = []; |
|
125
|
|
|
for ($j=0; $j<$ndim; $j++) { |
|
126
|
|
|
$npoint = array_merge($npoint, $transformer[$j]($this->data[$i][$j])); |
|
127
|
|
|
} |
|
128
|
|
|
$ndata[] = $npoint; |
|
129
|
|
|
} |
|
130
|
|
|
return $ndata; |
|
131
|
|
|
} |
|
132
|
|
|
|
|
133
|
|
|
/* |
|
134
|
|
|
* @param |
|
135
|
|
|
*/ |
|
136
|
|
|
function getLabelsAsArray() |
|
|
|
|
|
|
137
|
|
|
{ |
|
138
|
|
|
$resp = []; |
|
139
|
|
|
$len = count($data[0]); |
|
|
|
|
|
|
140
|
|
|
for ($i=0; $i<$len; $i++) { |
|
141
|
|
|
if (isset($this->encodedValues[$i])) { |
|
142
|
|
|
$resp = array_merge($resp, $this->encodedValues[$i]); |
|
143
|
|
|
} else { |
|
144
|
|
|
$resp[] = ""; |
|
145
|
|
|
} |
|
146
|
|
|
} |
|
147
|
|
|
return $resp; |
|
148
|
|
|
} |
|
149
|
|
|
|
|
150
|
|
|
/* |
|
151
|
|
|
* @param |
|
152
|
|
|
* |
|
153
|
|
|
function decode($ndata){ |
|
154
|
|
|
$resp = []; |
|
155
|
|
|
foreach ($ndata as $row) { |
|
156
|
|
|
$resp[] = $this->decodeRow($row); |
|
157
|
|
|
} |
|
158
|
|
|
return $resp; |
|
159
|
|
|
} |
|
160
|
|
|
|
|
161
|
|
|
/* |
|
162
|
|
|
* @param |
|
163
|
|
|
* |
|
164
|
|
|
function decodeRow($row){ |
|
165
|
|
|
$resp = []; |
|
166
|
|
|
$n = count($row); |
|
167
|
|
|
for ($i=0; $i<$n; $i++) { |
|
168
|
|
|
$resp[] = $this->decodeFeature($i, $row); |
|
169
|
|
|
} |
|
170
|
|
|
return $resp; |
|
171
|
|
|
}*/ |
|
172
|
|
|
|
|
173
|
|
|
} |
|
174
|
|
|
|
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.