|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace Phpml\SupportVectorMachine; |
|
6
|
|
|
|
|
7
|
|
|
class DataTransformer |
|
8
|
|
|
{ |
|
9
|
|
|
public static function trainingSet(array $samples, array $labels, bool $targets = false): string |
|
10
|
|
|
{ |
|
11
|
|
|
$set = ''; |
|
12
|
|
|
$numericLabels = []; |
|
13
|
|
|
|
|
14
|
|
|
if (!$targets) { |
|
15
|
|
|
$numericLabels = self::numericLabels($labels); |
|
16
|
|
|
} |
|
17
|
|
|
|
|
18
|
|
|
foreach ($labels as $index => $label) { |
|
19
|
|
|
$set .= sprintf('%s %s %s', ($targets ? $label : $numericLabels[$label]), self::sampleRow($samples[$index]), PHP_EOL); |
|
20
|
|
|
} |
|
21
|
|
|
|
|
22
|
|
|
return $set; |
|
23
|
|
|
} |
|
24
|
|
|
|
|
25
|
|
|
public static function testSet(array $samples): string |
|
26
|
|
|
{ |
|
27
|
|
|
if (!is_array($samples[0])) { |
|
28
|
|
|
$samples = [$samples]; |
|
29
|
|
|
} |
|
30
|
|
|
|
|
31
|
|
|
$set = ''; |
|
32
|
|
|
foreach ($samples as $sample) { |
|
33
|
|
|
$set .= sprintf('0 %s %s', self::sampleRow($sample), PHP_EOL); |
|
34
|
|
|
} |
|
35
|
|
|
|
|
36
|
|
|
return $set; |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
public static function predictions(string $rawPredictions, array $labels): array |
|
40
|
|
|
{ |
|
41
|
|
|
$numericLabels = self::numericLabels($labels); |
|
42
|
|
|
$results = []; |
|
43
|
|
|
foreach (explode(PHP_EOL, $rawPredictions) as $result) { |
|
44
|
|
|
if (isset($result[0])) { |
|
45
|
|
|
$results[] = array_search($result, $numericLabels); |
|
46
|
|
|
} |
|
47
|
|
|
} |
|
48
|
|
|
|
|
49
|
|
|
return $results; |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
public static function probabilities(string $rawPredictions, array $labels): array |
|
53
|
|
|
{ |
|
54
|
|
|
$numericLabels = self::numericLabels($labels); |
|
55
|
|
|
|
|
56
|
|
|
$predictions = explode(PHP_EOL, trim($rawPredictions)); |
|
57
|
|
|
|
|
58
|
|
|
$header = array_shift($predictions); |
|
59
|
|
|
$headerColumns = explode(' ', $header); |
|
60
|
|
|
array_shift($headerColumns); |
|
61
|
|
|
|
|
62
|
|
|
$columnLabels = []; |
|
63
|
|
|
foreach ($headerColumns as $numericLabel) { |
|
64
|
|
|
$columnLabels[] = array_search($numericLabel, $numericLabels); |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
|
|
$results = []; |
|
68
|
|
|
foreach ($predictions as $rawResult) { |
|
69
|
|
|
$probabilities = explode(' ', $rawResult); |
|
70
|
|
|
array_shift($probabilities); |
|
71
|
|
|
|
|
72
|
|
|
$result = []; |
|
73
|
|
|
foreach ($probabilities as $i => $prob) { |
|
74
|
|
|
$result[$columnLabels[$i]] = (float) $prob; |
|
75
|
|
|
} |
|
76
|
|
|
|
|
77
|
|
|
$results[] = $result; |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
return $results; |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
public static function numericLabels(array $labels): array |
|
84
|
|
|
{ |
|
85
|
|
|
$numericLabels = []; |
|
86
|
|
|
foreach ($labels as $label) { |
|
87
|
|
|
if (isset($numericLabels[$label])) { |
|
88
|
|
|
continue; |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
$numericLabels[$label] = count($numericLabels); |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
return $numericLabels; |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
private static function sampleRow(array $sample): string |
|
98
|
|
|
{ |
|
99
|
|
|
$row = []; |
|
100
|
|
|
foreach ($sample as $index => $feature) { |
|
101
|
|
|
$row[] = sprintf('%s:%s', $index + 1, $feature); |
|
102
|
|
|
} |
|
103
|
|
|
|
|
104
|
|
|
return implode(' ', $row); |
|
105
|
|
|
} |
|
106
|
|
|
} |
|
107
|
|
|
|
php-ai /
php-ml
