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