|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace devfym\IntelliPHP\Math; |
|
4
|
|
|
|
|
5
|
|
|
use devfym\IntelliPHP\Data\DataFrame; |
|
6
|
|
|
|
|
7
|
|
|
class Correlation |
|
8
|
|
|
{ |
|
9
|
|
|
/** |
|
10
|
|
|
* @param $df |
|
11
|
|
|
* @param $xColumn |
|
12
|
|
|
* @param $yColumn |
|
13
|
|
|
* @return float |
|
14
|
|
|
* Get Pearson's Correlation Coefficient. |
|
15
|
|
|
*/ |
|
16
|
|
|
public static function pearson(DataFrame $df, $xColumn, $yColumn) : float |
|
17
|
|
|
{ |
|
18
|
|
|
$n = $df->getIndex(); |
|
19
|
|
|
$x = 0; $y = 0; $xy = 0; $x2 = 0; $y2 = 0; |
|
20
|
|
|
|
|
21
|
|
|
for ($i = 0; $i < $n; $i++) { |
|
22
|
|
|
|
|
23
|
|
|
$cx = $df->{$xColumn}->get($i); |
|
24
|
|
|
$cy = $df->{$yColumn}->get($i); |
|
25
|
|
|
|
|
26
|
|
|
$x += $cx; |
|
27
|
|
|
$y += $cy; |
|
28
|
|
|
|
|
29
|
|
|
$xy += $cx * $cy; |
|
30
|
|
|
|
|
31
|
|
|
$x2 += $cx * $cx; |
|
32
|
|
|
$y2 += $cy * $cy; |
|
33
|
|
|
|
|
34
|
|
|
} |
|
35
|
|
|
|
|
36
|
|
|
$r = (($n * $xy) - ($x * $y)) / sqrt((($n * $x2) - ($x * $x)) * (($n * $y2) - ($y * $y))); |
|
37
|
|
|
|
|
38
|
|
|
return round($r, 4); |
|
39
|
|
|
} |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* @param DataFrame $df |
|
43
|
|
|
* @param $xColumn |
|
44
|
|
|
* @param $yColumn |
|
45
|
|
|
* @return float |
|
46
|
|
|
*/ |
|
47
|
|
|
public static function spearman(DataFrame $df, $xColumn, $yColumn) : float |
|
48
|
|
|
{ |
|
49
|
|
|
$xValue = $df->{$xColumn}->all(); |
|
50
|
|
|
$yValue = $df->{$yColumn}->all(); |
|
51
|
|
|
|
|
52
|
|
|
$xSort = array_unique($xValue); |
|
53
|
|
|
$ySort = array_unique($yValue); |
|
54
|
|
|
|
|
55
|
|
|
rsort($xSort); |
|
56
|
|
|
rsort($ySort); |
|
57
|
|
|
|
|
58
|
|
|
$xRank = []; |
|
59
|
|
|
$yRank = []; |
|
60
|
|
|
$diffRank = 0; |
|
61
|
|
|
|
|
62
|
|
|
for ($i = 0; $i < $df->getIndex(); $i++) { |
|
63
|
|
|
|
|
64
|
|
|
$xR = array_keys($xSort, $xValue[$i]); |
|
65
|
|
|
$yR = array_keys($ySort, $yValue[$i]); |
|
66
|
|
|
|
|
67
|
|
|
$xRank[$i] = $xR[0] + 1; |
|
68
|
|
|
$yRank[$i] = $yR[0] + 1; |
|
69
|
|
|
|
|
70
|
|
|
$diffRank += pow($xRank[$i] - $yRank[$i], 2); |
|
71
|
|
|
|
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
$p = 1 - ((6 * $diffRank) / ($df->getIndex() * (($df->getIndex() * $df->getIndex()) - 1))); |
|
75
|
|
|
|
|
76
|
|
|
return round($p, 4); |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
/** |
|
80
|
|
|
* @param DataFrame $df |
|
81
|
|
|
* @param $xColumn |
|
82
|
|
|
* @return float |
|
83
|
|
|
*/ |
|
84
|
|
|
public static function kendall(DataFrame $df, $xColumn) : float |
|
85
|
|
|
{ |
|
86
|
|
|
|
|
87
|
|
|
$concordant = []; |
|
88
|
|
|
$discordant = []; |
|
89
|
|
|
|
|
90
|
|
|
for ($i = 0; $i < $df->getIndex(); $i++) { |
|
91
|
|
|
|
|
92
|
|
|
$concordant_count = 0; |
|
93
|
|
|
$discordant_count = 0; |
|
94
|
|
|
|
|
95
|
|
|
for ($j = $i + 1; $j < $df->getIndex(); $j++) { |
|
96
|
|
|
|
|
97
|
|
|
if ($df->{$xColumn}->get($i) < $df->{$xColumn}->get($j)) { |
|
98
|
|
|
|
|
99
|
|
|
$concordant_count++; |
|
100
|
|
|
|
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
|
|
if ($df->{$xColumn}->get($i) > $df->{$xColumn}->get($j)) { |
|
104
|
|
|
|
|
105
|
|
|
$discordant_count++; |
|
106
|
|
|
|
|
107
|
|
|
} |
|
108
|
|
|
} |
|
109
|
|
|
|
|
110
|
|
|
$concordant[$i] = $concordant_count; |
|
111
|
|
|
$discordant[$i] = $discordant_count; |
|
112
|
|
|
|
|
113
|
|
|
} |
|
114
|
|
|
|
|
115
|
|
|
$scon = array_sum($concordant); |
|
116
|
|
|
$sdis = array_sum($discordant); |
|
117
|
|
|
|
|
118
|
|
|
$t = ($scon - $sdis) / ($df->getIndex() * ($df->getIndex() - 1) / 2); |
|
119
|
|
|
|
|
120
|
|
|
return round($t, 4); |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
/** |
|
124
|
|
|
* @param DataFrame $df |
|
125
|
|
|
* @return array |
|
126
|
|
|
*/ |
|
127
|
|
|
public static function pearsonAll(DataFrame $df) : array |
|
128
|
|
|
{ |
|
129
|
|
|
$arr = []; |
|
130
|
|
|
|
|
131
|
|
|
$columns = $df->getNumericColumns(); |
|
132
|
|
|
|
|
133
|
|
|
$numeric_count = count($columns); |
|
134
|
|
|
|
|
135
|
|
|
for ($i = 0; $i < $numeric_count; $i++) { |
|
136
|
|
|
|
|
137
|
|
|
for ($j = 0; $j < $numeric_count; $j++) { |
|
138
|
|
|
|
|
139
|
|
|
$arr[$i][$j] = self::pearson($df, $columns[$i], $columns[$j]); |
|
140
|
|
|
|
|
141
|
|
|
} |
|
142
|
|
|
|
|
143
|
|
|
} |
|
144
|
|
|
|
|
145
|
|
|
return $arr; |
|
146
|
|
|
} |
|
147
|
|
|
|
|
148
|
|
|
/** |
|
149
|
|
|
* @param DataFrame $df |
|
150
|
|
|
* @return array |
|
151
|
|
|
*/ |
|
152
|
|
|
public static function spearmanAll(DataFrame $df) : array |
|
153
|
|
|
{ |
|
154
|
|
|
$arr = []; |
|
155
|
|
|
|
|
156
|
|
|
$columns = $df->getNumericColumns(); |
|
157
|
|
|
|
|
158
|
|
|
$numeric_count = count($columns); |
|
159
|
|
|
|
|
160
|
|
|
for ($i = 0; $i < $numeric_count; $i++) { |
|
161
|
|
|
|
|
162
|
|
|
for ($j = 0; $j < $numeric_count; $j++) { |
|
163
|
|
|
|
|
164
|
|
|
$arr[$i][$j] = self::spearman($df, $columns[$i], $columns[$j]); |
|
165
|
|
|
|
|
166
|
|
|
} |
|
167
|
|
|
|
|
168
|
|
|
} |
|
169
|
|
|
|
|
170
|
|
|
return $arr; |
|
171
|
|
|
} |
|
172
|
|
|
} |
devfym /
intelliphp
