1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace PhpOffice\PhpSpreadsheet\Shared\Trend; |
4
|
|
|
|
5
|
|
|
abstract class BestFit |
6
|
|
|
{ |
7
|
|
|
/** |
8
|
|
|
* Indicator flag for a calculation error. |
9
|
|
|
* |
10
|
|
|
* @var bool |
11
|
|
|
*/ |
12
|
|
|
protected $error = false; |
13
|
|
|
|
14
|
|
|
/** |
15
|
|
|
* Algorithm type to use for best-fit. |
16
|
|
|
* |
17
|
|
|
* @var string |
18
|
|
|
*/ |
19
|
|
|
protected $bestFitType = 'undetermined'; |
20
|
|
|
|
21
|
|
|
/** |
22
|
|
|
* Number of entries in the sets of x- and y-value arrays. |
23
|
|
|
*/ |
24
|
|
|
protected int $valueCount; |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* X-value dataseries of values. |
28
|
|
|
* |
29
|
|
|
* @var float[] |
30
|
|
|
*/ |
31
|
|
|
protected $xValues = []; |
32
|
|
|
|
33
|
|
|
/** |
34
|
|
|
* Y-value dataseries of values. |
35
|
|
|
* |
36
|
|
|
* @var float[] |
37
|
|
|
*/ |
38
|
|
|
protected $yValues = []; |
39
|
|
|
|
40
|
|
|
/** |
41
|
|
|
* Flag indicating whether values should be adjusted to Y=0. |
42
|
|
|
* |
43
|
|
|
* @var bool |
44
|
|
|
*/ |
45
|
|
|
protected $adjustToZero = false; |
46
|
|
|
|
47
|
|
|
/** |
48
|
|
|
* Y-value series of best-fit values. |
49
|
|
|
* |
50
|
|
|
* @var float[] |
51
|
|
|
*/ |
52
|
|
|
protected $yBestFitValues = []; |
53
|
|
|
|
54
|
|
|
/** @var float */ |
55
|
|
|
protected $goodnessOfFit = 1; |
56
|
|
|
|
57
|
|
|
/** @var float */ |
58
|
|
|
protected $stdevOfResiduals = 0; |
59
|
|
|
|
60
|
|
|
/** @var float */ |
61
|
|
|
protected $covariance = 0; |
62
|
|
|
|
63
|
|
|
/** @var float */ |
64
|
|
|
protected $correlation = 0; |
65
|
|
|
|
66
|
|
|
/** @var float */ |
67
|
|
|
protected $SSRegression = 0; |
68
|
|
|
|
69
|
|
|
/** @var float */ |
70
|
|
|
protected $SSResiduals = 0; |
71
|
|
|
|
72
|
|
|
/** @var float */ |
73
|
|
|
protected $DFResiduals = 0; |
74
|
|
|
|
75
|
|
|
/** @var float */ |
76
|
|
|
protected $f = 0; |
77
|
|
|
|
78
|
|
|
/** @var float */ |
79
|
|
|
protected $slope = 0; |
80
|
|
|
|
81
|
|
|
/** @var float */ |
82
|
|
|
protected $slopeSE = 0; |
83
|
|
|
|
84
|
|
|
/** @var float */ |
85
|
|
|
protected $intersect = 0; |
86
|
|
|
|
87
|
|
|
/** @var float */ |
88
|
|
|
protected $intersectSE = 0; |
89
|
|
|
|
90
|
|
|
/** @var float */ |
91
|
|
|
protected $xOffset = 0; |
92
|
|
|
|
93
|
|
|
/** @var float */ |
94
|
|
|
protected $yOffset = 0; |
95
|
|
|
|
96
|
|
|
/** @return bool */ |
97
|
|
|
public function getError() |
98
|
|
|
{ |
99
|
|
|
return $this->error; |
100
|
|
|
} |
101
|
|
|
|
102
|
|
|
/** @return string */ |
103
|
|
|
public function getBestFitType() |
104
|
|
|
{ |
105
|
|
|
return $this->bestFitType; |
106
|
|
|
} |
107
|
|
|
|
108
|
|
|
/** |
109
|
|
|
* Return the Y-Value for a specified value of X. |
110
|
|
|
* |
111
|
|
|
* @param float $xValue X-Value |
112
|
|
|
* |
113
|
|
|
* @return float Y-Value |
114
|
|
|
*/ |
115
|
|
|
abstract public function getValueOfYForX($xValue); |
116
|
|
|
|
117
|
|
|
/** |
118
|
|
|
* Return the X-Value for a specified value of Y. |
119
|
|
|
* |
120
|
|
|
* @param float $yValue Y-Value |
121
|
|
|
* |
122
|
|
|
* @return float X-Value |
123
|
|
|
*/ |
124
|
|
|
abstract public function getValueOfXForY($yValue); |
125
|
|
|
|
126
|
|
|
/** |
127
|
|
|
* Return the original set of X-Values. |
128
|
|
|
* |
129
|
|
|
* @return float[] X-Values |
130
|
|
|
*/ |
131
|
4 |
|
public function getXValues() |
132
|
|
|
{ |
133
|
4 |
|
return $this->xValues; |
134
|
|
|
} |
135
|
|
|
|
136
|
|
|
/** |
137
|
|
|
* Return the Equation of the best-fit line. |
138
|
|
|
* |
139
|
|
|
* @param int $dp Number of places of decimal precision to display |
140
|
|
|
* |
141
|
|
|
* @return string |
142
|
|
|
*/ |
143
|
|
|
abstract public function getEquation($dp = 0); |
144
|
|
|
|
145
|
|
|
/** |
146
|
|
|
* Return the Slope of the line. |
147
|
|
|
* |
148
|
|
|
* @param int $dp Number of places of decimal precision to display |
149
|
|
|
* |
150
|
|
|
* @return float |
151
|
|
|
*/ |
152
|
41 |
|
public function getSlope($dp = 0) |
153
|
|
|
{ |
154
|
41 |
|
if ($dp != 0) { |
155
|
2 |
|
return round($this->slope, $dp); |
156
|
|
|
} |
157
|
|
|
|
158
|
41 |
|
return $this->slope; |
159
|
|
|
} |
160
|
|
|
|
161
|
|
|
/** |
162
|
|
|
* Return the standard error of the Slope. |
163
|
|
|
* |
164
|
|
|
* @param int $dp Number of places of decimal precision to display |
165
|
|
|
* |
166
|
|
|
* @return float |
167
|
|
|
*/ |
168
|
3 |
|
public function getSlopeSE($dp = 0) |
169
|
|
|
{ |
170
|
3 |
|
if ($dp != 0) { |
171
|
|
|
return round($this->slopeSE, $dp); |
172
|
|
|
} |
173
|
|
|
|
174
|
3 |
|
return $this->slopeSE; |
175
|
|
|
} |
176
|
|
|
|
177
|
|
|
/** |
178
|
|
|
* Return the Value of X where it intersects Y = 0. |
179
|
|
|
* |
180
|
|
|
* @param int $dp Number of places of decimal precision to display |
181
|
|
|
* |
182
|
|
|
* @return float |
183
|
|
|
*/ |
184
|
39 |
|
public function getIntersect($dp = 0) |
185
|
|
|
{ |
186
|
39 |
|
if ($dp != 0) { |
187
|
2 |
|
return round($this->intersect, $dp); |
188
|
|
|
} |
189
|
|
|
|
190
|
39 |
|
return $this->intersect; |
191
|
|
|
} |
192
|
|
|
|
193
|
|
|
/** |
194
|
|
|
* Return the standard error of the Intersect. |
195
|
|
|
* |
196
|
|
|
* @param int $dp Number of places of decimal precision to display |
197
|
|
|
* |
198
|
|
|
* @return float |
199
|
|
|
*/ |
200
|
2 |
|
public function getIntersectSE($dp = 0) |
201
|
|
|
{ |
202
|
2 |
|
if ($dp != 0) { |
203
|
|
|
return round($this->intersectSE, $dp); |
204
|
|
|
} |
205
|
|
|
|
206
|
2 |
|
return $this->intersectSE; |
207
|
|
|
} |
208
|
|
|
|
209
|
|
|
/** |
210
|
|
|
* Return the goodness of fit for this regression. |
211
|
|
|
* |
212
|
|
|
* @param int $dp Number of places of decimal precision to return |
213
|
|
|
* |
214
|
|
|
* @return float |
215
|
|
|
*/ |
216
|
10 |
|
public function getGoodnessOfFit($dp = 0) |
217
|
|
|
{ |
218
|
10 |
|
if ($dp != 0) { |
219
|
3 |
|
return round($this->goodnessOfFit, $dp); |
220
|
|
|
} |
221
|
|
|
|
222
|
10 |
|
return $this->goodnessOfFit; |
223
|
|
|
} |
224
|
|
|
|
225
|
|
|
/** |
226
|
|
|
* Return the goodness of fit for this regression. |
227
|
|
|
* |
228
|
|
|
* @param int $dp Number of places of decimal precision to return |
229
|
|
|
* |
230
|
|
|
* @return float |
231
|
|
|
*/ |
232
|
|
|
public function getGoodnessOfFitPercent($dp = 0) |
233
|
|
|
{ |
234
|
|
|
if ($dp != 0) { |
235
|
|
|
return round($this->goodnessOfFit * 100, $dp); |
236
|
|
|
} |
237
|
|
|
|
238
|
|
|
return $this->goodnessOfFit * 100; |
239
|
|
|
} |
240
|
|
|
|
241
|
|
|
/** |
242
|
|
|
* Return the standard deviation of the residuals for this regression. |
243
|
|
|
* |
244
|
|
|
* @param int $dp Number of places of decimal precision to return |
245
|
|
|
* |
246
|
|
|
* @return float |
247
|
|
|
*/ |
248
|
7 |
|
public function getStdevOfResiduals($dp = 0) |
249
|
|
|
{ |
250
|
7 |
|
if ($dp != 0) { |
251
|
|
|
return round($this->stdevOfResiduals, $dp); |
252
|
|
|
} |
253
|
|
|
|
254
|
7 |
|
return $this->stdevOfResiduals; |
255
|
|
|
} |
256
|
|
|
|
257
|
|
|
/** |
258
|
|
|
* @param int $dp Number of places of decimal precision to return |
259
|
|
|
* |
260
|
|
|
* @return float |
261
|
|
|
*/ |
262
|
3 |
|
public function getSSRegression($dp = 0) |
263
|
|
|
{ |
264
|
3 |
|
if ($dp != 0) { |
265
|
|
|
return round($this->SSRegression, $dp); |
266
|
|
|
} |
267
|
|
|
|
268
|
3 |
|
return $this->SSRegression; |
269
|
|
|
} |
270
|
|
|
|
271
|
|
|
/** |
272
|
|
|
* @param int $dp Number of places of decimal precision to return |
273
|
|
|
* |
274
|
|
|
* @return float |
275
|
|
|
*/ |
276
|
3 |
|
public function getSSResiduals($dp = 0) |
277
|
|
|
{ |
278
|
3 |
|
if ($dp != 0) { |
279
|
|
|
return round($this->SSResiduals, $dp); |
280
|
|
|
} |
281
|
|
|
|
282
|
3 |
|
return $this->SSResiduals; |
283
|
|
|
} |
284
|
|
|
|
285
|
|
|
/** |
286
|
|
|
* @param int $dp Number of places of decimal precision to return |
287
|
|
|
* |
288
|
|
|
* @return float |
289
|
|
|
*/ |
290
|
3 |
|
public function getDFResiduals($dp = 0) |
291
|
|
|
{ |
292
|
3 |
|
if ($dp != 0) { |
293
|
|
|
return round($this->DFResiduals, $dp); |
294
|
|
|
} |
295
|
|
|
|
296
|
3 |
|
return $this->DFResiduals; |
297
|
|
|
} |
298
|
|
|
|
299
|
|
|
/** |
300
|
|
|
* @param int $dp Number of places of decimal precision to return |
301
|
|
|
* |
302
|
|
|
* @return float |
303
|
|
|
*/ |
304
|
3 |
|
public function getF($dp = 0) |
305
|
|
|
{ |
306
|
3 |
|
if ($dp != 0) { |
307
|
|
|
return round($this->f, $dp); |
308
|
|
|
} |
309
|
|
|
|
310
|
3 |
|
return $this->f; |
311
|
|
|
} |
312
|
|
|
|
313
|
|
|
/** |
314
|
|
|
* @param int $dp Number of places of decimal precision to return |
315
|
|
|
* |
316
|
|
|
* @return float |
317
|
|
|
*/ |
318
|
5 |
|
public function getCovariance($dp = 0) |
319
|
|
|
{ |
320
|
5 |
|
if ($dp != 0) { |
321
|
|
|
return round($this->covariance, $dp); |
322
|
|
|
} |
323
|
|
|
|
324
|
5 |
|
return $this->covariance; |
325
|
|
|
} |
326
|
|
|
|
327
|
|
|
/** |
328
|
|
|
* @param int $dp Number of places of decimal precision to return |
329
|
|
|
* |
330
|
|
|
* @return float |
331
|
|
|
*/ |
332
|
4 |
|
public function getCorrelation($dp = 0) |
333
|
|
|
{ |
334
|
4 |
|
if ($dp != 0) { |
335
|
|
|
return round($this->correlation, $dp); |
336
|
|
|
} |
337
|
|
|
|
338
|
4 |
|
return $this->correlation; |
339
|
|
|
} |
340
|
|
|
|
341
|
|
|
/** |
342
|
|
|
* @return float[] |
343
|
|
|
*/ |
344
|
|
|
public function getYBestFitValues() |
345
|
|
|
{ |
346
|
|
|
return $this->yBestFitValues; |
347
|
|
|
} |
348
|
|
|
|
349
|
|
|
/** |
350
|
|
|
* @param float $sumX |
351
|
|
|
* @param float $sumY |
352
|
37 |
|
* @param float $sumX2 |
353
|
|
|
* @param float $sumY2 |
354
|
37 |
|
* @param float $sumXY |
355
|
|
|
* @param float $meanX |
356
|
|
|
* @param float $meanY |
357
|
|
|
* @param bool|int $const |
358
|
|
|
*/ |
359
|
|
|
protected function calculateGoodnessOfFit($sumX, $sumY, $sumX2, $sumY2, $sumXY, $meanX, $meanY, $const): void |
360
|
|
|
{ |
361
|
|
|
$SSres = $SScov = $SStot = $SSsex = 0.0; |
362
|
|
|
foreach ($this->xValues as $xKey => $xValue) { |
363
|
|
|
$bestFitY = $this->yBestFitValues[$xKey] = $this->getValueOfYForX($xValue); |
364
|
|
|
|
365
|
|
|
$SSres += ($this->yValues[$xKey] - $bestFitY) * ($this->yValues[$xKey] - $bestFitY); |
366
|
|
|
if ($const === true) { |
367
|
37 |
|
$SStot += ($this->yValues[$xKey] - $meanY) * ($this->yValues[$xKey] - $meanY); |
368
|
|
|
} else { |
369
|
37 |
|
$SStot += $this->yValues[$xKey] * $this->yValues[$xKey]; |
370
|
37 |
|
} |
371
|
37 |
|
$SScov += ($this->xValues[$xKey] - $meanX) * ($this->yValues[$xKey] - $meanY); |
372
|
|
|
if ($const === true) { |
373
|
37 |
|
$SSsex += ($this->xValues[$xKey] - $meanX) * ($this->xValues[$xKey] - $meanX); |
374
|
37 |
|
} else { |
375
|
32 |
|
$SSsex += $this->xValues[$xKey] * $this->xValues[$xKey]; |
376
|
|
|
} |
377
|
5 |
|
} |
378
|
|
|
|
379
|
37 |
|
$this->SSResiduals = $SSres; |
380
|
37 |
|
$this->DFResiduals = $this->valueCount - 1 - ($const === true ? 1 : 0); |
381
|
32 |
|
|
382
|
|
|
if ($this->DFResiduals == 0.0) { |
383
|
5 |
|
$this->stdevOfResiduals = 0.0; |
384
|
|
|
} else { |
385
|
|
|
$this->stdevOfResiduals = sqrt($SSres / $this->DFResiduals); |
386
|
|
|
} |
387
|
37 |
|
|
388
|
37 |
|
if ($SStot == 0.0 || $SSres == $SStot) { |
|
|
|
|
389
|
|
|
$this->goodnessOfFit = 1; |
390
|
37 |
|
} else { |
391
|
1 |
|
$this->goodnessOfFit = 1 - ($SSres / $SStot); |
392
|
|
|
} |
393
|
36 |
|
|
394
|
|
|
$this->SSRegression = $this->goodnessOfFit * $SStot; |
395
|
|
|
$this->covariance = $SScov / $this->valueCount; |
396
|
37 |
|
$this->correlation = ($this->valueCount * $sumXY - $sumX * $sumY) / sqrt(($this->valueCount * $sumX2 - $sumX ** 2) * ($this->valueCount * $sumY2 - $sumY ** 2)); |
397
|
|
|
$this->slopeSE = $this->stdevOfResiduals / sqrt($SSsex); |
398
|
|
|
$this->intersectSE = $this->stdevOfResiduals * sqrt(1 / ($this->valueCount - ($sumX * $sumX) / $sumX2)); |
399
|
37 |
|
if ($this->SSResiduals != 0.0) { |
400
|
|
|
if ($this->DFResiduals == 0.0) { |
401
|
|
|
$this->f = 0.0; |
402
|
37 |
|
} else { |
403
|
37 |
|
$this->f = $this->SSRegression / ($this->SSResiduals / $this->DFResiduals); |
404
|
37 |
|
} |
405
|
37 |
|
} else { |
406
|
37 |
|
if ($this->DFResiduals == 0.0) { |
407
|
37 |
|
$this->f = 0.0; |
408
|
26 |
|
} else { |
409
|
|
|
$this->f = $this->SSRegression / $this->DFResiduals; |
410
|
|
|
} |
411
|
26 |
|
} |
412
|
|
|
} |
413
|
|
|
|
414
|
12 |
|
/** @return float|int */ |
415
|
1 |
|
private function sumSquares(array $values) |
416
|
|
|
{ |
417
|
11 |
|
return array_sum( |
418
|
|
|
array_map( |
419
|
|
|
fn ($value): float|int => $value ** 2, |
420
|
|
|
$values |
421
|
|
|
) |
422
|
|
|
); |
423
|
37 |
|
} |
424
|
|
|
|
425
|
37 |
|
/** |
426
|
37 |
|
* @param float[] $yValues |
427
|
37 |
|
* @param float[] $xValues |
428
|
37 |
|
*/ |
429
|
37 |
|
protected function leastSquareFit(array $yValues, array $xValues, bool $const): void |
430
|
37 |
|
{ |
431
|
|
|
// calculate sums |
432
|
|
|
$sumValuesX = array_sum($xValues); |
433
|
|
|
$sumValuesY = array_sum($yValues); |
434
|
|
|
$meanValueX = $sumValuesX / $this->valueCount; |
435
|
|
|
$meanValueY = $sumValuesY / $this->valueCount; |
436
|
|
|
$sumSquaresX = $this->sumSquares($xValues); |
437
|
37 |
|
$sumSquaresY = $this->sumSquares($yValues); |
438
|
|
|
$mBase = $mDivisor = 0.0; |
439
|
|
|
$xy_sum = 0.0; |
440
|
37 |
|
for ($i = 0; $i < $this->valueCount; ++$i) { |
441
|
37 |
|
$xy_sum += $xValues[$i] * $yValues[$i]; |
442
|
37 |
|
|
443
|
37 |
|
if ($const === true) { |
444
|
37 |
|
$mBase += ($xValues[$i] - $meanValueX) * ($yValues[$i] - $meanValueY); |
445
|
37 |
|
$mDivisor += ($xValues[$i] - $meanValueX) * ($xValues[$i] - $meanValueX); |
446
|
37 |
|
} else { |
447
|
37 |
|
$mBase += $xValues[$i] * $yValues[$i]; |
448
|
37 |
|
$mDivisor += $xValues[$i] * $xValues[$i]; |
449
|
37 |
|
} |
450
|
|
|
} |
451
|
37 |
|
|
452
|
32 |
|
// calculate slope |
453
|
32 |
|
$this->slope = $mBase / $mDivisor; |
454
|
|
|
|
455
|
5 |
|
// calculate intersect |
456
|
5 |
|
$this->intersect = ($const === true) ? $meanValueY - ($this->slope * $meanValueX) : 0.0; |
457
|
|
|
|
458
|
|
|
$this->calculateGoodnessOfFit($sumValuesX, $sumValuesY, $sumSquaresX, $sumSquaresY, $xy_sum, $meanValueX, $meanValueY, $const); |
459
|
|
|
} |
460
|
|
|
|
461
|
37 |
|
/** |
462
|
|
|
* Define the regression. |
463
|
|
|
* |
464
|
37 |
|
* @param float[] $yValues The set of Y-values for this regression |
465
|
|
|
* @param float[] $xValues The set of X-values for this regression |
466
|
37 |
|
*/ |
467
|
|
|
public function __construct($yValues, $xValues = []) |
468
|
|
|
{ |
469
|
|
|
// Calculate number of points |
470
|
|
|
$yValueCount = count($yValues); |
471
|
|
|
$xValueCount = count($xValues); |
472
|
|
|
|
473
|
|
|
// Define X Values if necessary |
474
|
|
|
if ($xValueCount === 0) { |
475
|
37 |
|
$xValues = range(1, $yValueCount); |
476
|
|
|
} elseif ($yValueCount !== $xValueCount) { |
477
|
|
|
// Ensure both arrays of points are the same size |
478
|
37 |
|
$this->error = true; |
479
|
37 |
|
} |
480
|
|
|
|
481
|
|
|
$this->valueCount = $yValueCount; |
482
|
37 |
|
$this->xValues = $xValues; |
483
|
|
|
$this->yValues = $yValues; |
484
|
37 |
|
} |
485
|
|
|
} |
486
|
|
|
|