|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace devfym\IntelliPHP\Data; |
|
4
|
|
|
|
|
5
|
|
|
use devfym\IntelliPHP\Math\StatisticInterface; |
|
6
|
|
|
|
|
7
|
|
|
class DataFrame implements StatisticInterface |
|
8
|
|
|
{ |
|
9
|
|
|
/** |
|
10
|
|
|
* @var array |
|
11
|
|
|
* List of Column names. |
|
12
|
|
|
*/ |
|
13
|
|
|
private $columns = []; |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* @var int |
|
17
|
|
|
* Number of Sample. |
|
18
|
|
|
*/ |
|
19
|
|
|
private $index = 0; |
|
20
|
|
|
|
|
21
|
|
|
/** |
|
22
|
|
|
* DataFrame constructor. |
|
23
|
|
|
*/ |
|
24
|
|
|
public function __construct() |
|
25
|
|
|
{ |
|
26
|
|
|
// |
|
27
|
|
|
} |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* @param array $arr |
|
31
|
|
|
* @return array |
|
32
|
|
|
* Transpose array. |
|
33
|
|
|
*/ |
|
34
|
|
|
public function transpose($arr = []) : array |
|
35
|
|
|
{ |
|
36
|
|
|
$columns = array_keys($arr[0]); |
|
37
|
|
|
|
|
38
|
|
|
$arr = array_map(null, ...$arr); |
|
39
|
|
|
|
|
40
|
|
|
foreach ($columns as $index => $column) { |
|
41
|
|
|
$arr[$column] = $arr[$index]; |
|
42
|
|
|
unset($arr[$index]); |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
|
|
return $arr; |
|
46
|
|
|
} |
|
47
|
|
|
|
|
48
|
|
|
/** |
|
49
|
|
|
* @param array $arr |
|
50
|
|
|
* @param bool $transpose |
|
51
|
|
|
* Store given array-formatted data into series. |
|
52
|
|
|
*/ |
|
53
|
|
|
public function readArray($arr = [], $transpose = false) : void |
|
54
|
|
|
{ |
|
55
|
|
|
|
|
56
|
|
|
if ($transpose === true) { |
|
57
|
|
|
$arr = $this->transpose($arr); |
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
|
|
// Set Columns |
|
61
|
|
|
$this->columns = array_keys($arr); |
|
62
|
|
|
|
|
63
|
|
|
// Set Index |
|
64
|
|
|
$this->index = count($arr[$this->columns[0]]); |
|
65
|
|
|
|
|
66
|
|
|
// Create Series instance for each column. |
|
67
|
|
|
foreach ($this->columns as $column) { |
|
68
|
|
|
$this->{$column} = new Series(); |
|
69
|
|
|
$this->{$column}->setSample($arr[$column]); |
|
70
|
|
|
} |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
/** |
|
74
|
|
|
* @return array |
|
75
|
|
|
* Get list of column names. |
|
76
|
|
|
*/ |
|
77
|
|
|
public function getColumns() : array |
|
78
|
|
|
{ |
|
79
|
|
|
return $this->columns; |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
|
|
public function getNumericColumns() : array { |
|
83
|
|
|
|
|
84
|
|
|
$columns = $this->columns; |
|
85
|
|
|
|
|
86
|
|
|
$numeric = []; |
|
87
|
|
|
|
|
88
|
|
|
foreach ($columns as $col) { |
|
89
|
|
|
if ($this->{$col}->dataType() == 'Numeric') { |
|
90
|
|
|
array_push($numeric, $col); |
|
91
|
|
|
} |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
return $numeric; |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
/** |
|
98
|
|
|
* @return int |
|
99
|
|
|
* Get Number of Sample. |
|
100
|
|
|
*/ |
|
101
|
|
|
public function getIndex() : int |
|
102
|
|
|
{ |
|
103
|
|
|
return $this->index; |
|
104
|
|
|
} |
|
105
|
|
|
|
|
106
|
|
|
/** |
|
107
|
|
|
* @param int $floatPoint |
|
108
|
|
|
* @return array |
|
109
|
|
|
* Get list of mean value in DataFrame. |
|
110
|
|
|
*/ |
|
111
|
|
|
public function mean($floatPoint = 1) : array |
|
112
|
|
|
{ |
|
113
|
|
|
// Initialize mean. |
|
114
|
|
|
$mean = []; |
|
115
|
|
|
|
|
116
|
|
|
foreach ($this->columns as $column) { |
|
117
|
|
|
if ($this->{$column}->dataType() == 'Numeric') { |
|
118
|
|
|
$mean[$column] = $this->{$column}->mean($floatPoint); |
|
119
|
|
|
} |
|
120
|
|
|
} |
|
121
|
|
|
|
|
122
|
|
|
return $mean; |
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
|
|
/** |
|
126
|
|
|
* @param int $floatPoint |
|
127
|
|
|
* @return array |
|
128
|
|
|
* Get list of max value in DataFrame. |
|
129
|
|
|
*/ |
|
130
|
|
|
public function max($floatPoint = 2) : array |
|
131
|
|
|
{ |
|
132
|
|
|
// Initialize max. |
|
133
|
|
|
$max = []; |
|
134
|
|
|
|
|
135
|
|
|
foreach ($this->columns as $column) { |
|
136
|
|
|
if ($this->{$column}->dataType() == 'Numeric') { |
|
137
|
|
|
$max[$column] = $this->{$column}->max($floatPoint); |
|
138
|
|
|
} |
|
139
|
|
|
} |
|
140
|
|
|
|
|
141
|
|
|
return $max; |
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
|
|
/** |
|
145
|
|
|
* @param int $floatPoint |
|
146
|
|
|
* @return array |
|
147
|
|
|
* Get list of min value in DataFrame. |
|
148
|
|
|
*/ |
|
149
|
|
|
public function min($floatPoint = 2) : array |
|
150
|
|
|
{ |
|
151
|
|
|
// Initialize min. |
|
152
|
|
|
$min = []; |
|
153
|
|
|
|
|
154
|
|
|
foreach ($this->columns as $column) { |
|
155
|
|
|
if ($this->{$column}->dataType() == 'Numeric') { |
|
156
|
|
|
$min[$column] = $this->{$column}->min($floatPoint); |
|
157
|
|
|
} |
|
158
|
|
|
} |
|
159
|
|
|
|
|
160
|
|
|
return $min; |
|
161
|
|
|
} |
|
162
|
|
|
|
|
163
|
|
|
/** |
|
164
|
|
|
* @param int $Q |
|
165
|
|
|
* @param int $floatPoint |
|
166
|
|
|
* @return array |
|
167
|
|
|
* Get list of Q-th quartile value in DataFrame. |
|
168
|
|
|
*/ |
|
169
|
|
|
public function quartile($Q = 1, $floatPoint = 2) : array |
|
170
|
|
|
{ |
|
171
|
|
|
// Initialize quartile. |
|
172
|
|
|
$quartile = []; |
|
173
|
|
|
|
|
174
|
|
|
foreach ($this->columns as $column) { |
|
175
|
|
|
if ($this->{$column}->dataType() == 'Numeric') { |
|
176
|
|
|
$quartile[$column] = $this->{$column}->quartile($Q, $floatPoint); |
|
177
|
|
|
} |
|
178
|
|
|
} |
|
179
|
|
|
|
|
180
|
|
|
return $quartile; |
|
181
|
|
|
} |
|
182
|
|
|
|
|
183
|
|
|
/** |
|
184
|
|
|
* @param int $floatPoint |
|
185
|
|
|
* @return array |
|
186
|
|
|
* Get list of 2nd Quartile value in DataFrame. |
|
187
|
|
|
*/ |
|
188
|
|
|
public function median($floatPoint = 2) : array |
|
189
|
|
|
{ |
|
190
|
|
|
return $this->quartile(2, $floatPoint); |
|
191
|
|
|
} |
|
192
|
|
|
|
|
193
|
|
|
/** |
|
194
|
|
|
* @param int $floatPoint |
|
195
|
|
|
* @return array |
|
196
|
|
|
* Get list of Variance in DataFrame. |
|
197
|
|
|
*/ |
|
198
|
|
|
public function variance($floatPoint = 4) : array |
|
199
|
|
|
{ |
|
200
|
|
|
// Initialize quartile. |
|
201
|
|
|
$variance = []; |
|
202
|
|
|
|
|
203
|
|
|
foreach ($this->columns as $column) { |
|
204
|
|
|
if ($this->{$column}->dataType() == 'Numeric') { |
|
205
|
|
|
$variance[$column] = $this->{$column}->variance($floatPoint); |
|
206
|
|
|
} |
|
207
|
|
|
} |
|
208
|
|
|
|
|
209
|
|
|
return $variance; |
|
210
|
|
|
} |
|
211
|
|
|
|
|
212
|
|
|
/** |
|
213
|
|
|
* @param int $floatPoint |
|
214
|
|
|
* @return array |
|
215
|
|
|
* Get list of Standard Deviation in DataFrame. |
|
216
|
|
|
*/ |
|
217
|
|
|
public function std($floatPoint = 2) : array |
|
218
|
|
|
{ |
|
219
|
|
|
// Initialize quartile. |
|
220
|
|
|
$std = []; |
|
221
|
|
|
|
|
222
|
|
|
foreach ($this->columns as $column) { |
|
223
|
|
|
if ($this->{$column}->dataType() == 'Numeric') { |
|
224
|
|
|
$std[$column] = $this->{$column}->std($floatPoint); |
|
225
|
|
|
} |
|
226
|
|
|
} |
|
227
|
|
|
|
|
228
|
|
|
return $std; |
|
229
|
|
|
} |
|
230
|
|
|
} |
|
231
|
|
|
|