Trend unit tests (#1899)

- Move TREND() functions into the Statistical Trends class - Unit tests for TREND() - Create Confidence class for Statistical Confidence functions
PHPOffice · Mar 6, 2021 · 2d8c8c8 · 2d8c8c8
1 parent a79a4dd
commit 2d8c8c8
Show file tree

Hide file tree

Showing 15 changed files with 344 additions and 111 deletions.
diff --git a/src/PhpSpreadsheet/Calculation/Statistical/Trends.php b/src/PhpSpreadsheet/Calculation/Statistical/Trends.php
@@ -226,7 +226,7 @@ public static function LINEST($yValues, $xValues = null, $const = true, $stats =
                 ],
                 [
                     $bestFitLinear->getSlopeSE(),
-                    $bestFitLinear->getIntersectSE(),
+                    ($const === false) ? Functions::NA() : $bestFitLinear->getIntersectSE(),
                 ],
                 [
                     $bestFitLinear->getGoodnessOfFit(),
@@ -293,7 +293,7 @@ public static function LOGEST($yValues, $xValues = null, $const = true, $stats =
                 ],
                 [
                     $bestFitExponential->getSlopeSE(),
-                    $bestFitExponential->getIntersectSE(),
+                    ($const === false) ? Functions::NA() : $bestFitExponential->getIntersectSE(),
                 ],
                 [
                     $bestFitExponential->getGoodnessOfFit(),

diff --git a/src/PhpSpreadsheet/Shared/Trend/BestFit.php b/src/PhpSpreadsheet/Shared/Trend/BestFit.php
@@ -348,21 +348,21 @@ protected function calculateGoodnessOfFit($sumX, $sumY, $sumX2, $sumY2, $sumXY,
             $bestFitY = $this->yBestFitValues[$xKey] = $this->getValueOfYForX($xValue);
 
             $SSres += ($this->yValues[$xKey] - $bestFitY) * ($this->yValues[$xKey] - $bestFitY);
-            if ($const) {
+            if ($const === true) {
                 $SStot += ($this->yValues[$xKey] - $meanY) * ($this->yValues[$xKey] - $meanY);
             } else {
                 $SStot += $this->yValues[$xKey] * $this->yValues[$xKey];
             }
             $SScov += ($this->xValues[$xKey] - $meanX) * ($this->yValues[$xKey] - $meanY);
-            if ($const) {
+            if ($const === true) {
                 $SSsex += ($this->xValues[$xKey] - $meanX) * ($this->xValues[$xKey] - $meanX);
             } else {
                 $SSsex += $this->xValues[$xKey] * $this->xValues[$xKey];
             }
         }
 
         $this->SSResiduals = $SSres;
-        $this->DFResiduals = $this->valueCount - 1 - $const;
+        $this->DFResiduals = $this->valueCount - 1 - ($const === true ? 1 : 0);
 
         if ($this->DFResiduals == 0.0) {
             $this->stdevOfResiduals = 0.0;
@@ -395,27 +395,39 @@ protected function calculateGoodnessOfFit($sumX, $sumY, $sumX2, $sumY2, $sumXY,
         }
     }
 
+    private function sumSquares(array $values)
+    {
+        return array_sum(
+            array_map(
+                function ($value) {
+                    return $value ** 2;
+                },
+                $values
+            )
+        );
+    }
+
     /**
      * @param float[] $yValues
      * @param float[] $xValues
-     * @param bool $const
      */
-    protected function leastSquareFit(array $yValues, array $xValues, $const): void
+    protected function leastSquareFit(array $yValues, array $xValues, bool $const): void
     {
         // calculate sums
-        $x_sum = array_sum($xValues);
-        $y_sum = array_sum($yValues);
-        $meanX = $x_sum / $this->valueCount;
-        $meanY = $y_sum / $this->valueCount;
-        $mBase = $mDivisor = $xx_sum = $xy_sum = $yy_sum = 0.0;
+        $sumValuesX = array_sum($xValues);
+        $sumValuesY = array_sum($yValues);
+        $meanValueX = $sumValuesX / $this->valueCount;
+        $meanValueY = $sumValuesY / $this->valueCount;
+        $sumSquaresX = $this->sumSquares($xValues);
+        $sumSquaresY = $this->sumSquares($yValues);
+        $mBase = $mDivisor = 0.0;
+        $xy_sum = 0.0;
         for ($i = 0; $i < $this->valueCount; ++$i) {
             $xy_sum += $xValues[$i] * $yValues[$i];
-            $xx_sum += $xValues[$i] * $xValues[$i];
-            $yy_sum += $yValues[$i] * $yValues[$i];
 
-            if ($const) {
-                $mBase += ($xValues[$i] - $meanX) * ($yValues[$i] - $meanY);
-                $mDivisor += ($xValues[$i] - $meanX) * ($xValues[$i] - $meanX);
+            if ($const === true) {
+                $mBase += ($xValues[$i] - $meanValueX) * ($yValues[$i] - $meanValueY);
+                $mDivisor += ($xValues[$i] - $meanValueX) * ($xValues[$i] - $meanValueX);
             } else {
                 $mBase += $xValues[$i] * $yValues[$i];
                 $mDivisor += $xValues[$i] * $xValues[$i];
@@ -426,37 +438,32 @@ protected function leastSquareFit(array $yValues, array $xValues, $const): void
         $this->slope = $mBase / $mDivisor;
 
         // calculate intersect
-        if ($const) {
-            $this->intersect = $meanY - ($this->slope * $meanX);
-        } else {
-            $this->intersect = 0;
-        }
+        $this->intersect = ($const === true) ? $meanValueY - ($this->slope * $meanValueX) : 0.0;
 
-        $this->calculateGoodnessOfFit($x_sum, $y_sum, $xx_sum, $yy_sum, $xy_sum, $meanX, $meanY, $const);
+        $this->calculateGoodnessOfFit($sumValuesX, $sumValuesY, $sumSquaresX, $sumSquaresY, $xy_sum, $meanValueX, $meanValueY, $const);
     }
 
     /**
      * Define the regression.
      *
      * @param float[] $yValues The set of Y-values for this regression
      * @param float[] $xValues The set of X-values for this regression
-     * @param bool $const
      */
-    public function __construct($yValues, $xValues = [], $const = true)
+    public function __construct($yValues, $xValues = [])
     {
         //    Calculate number of points
-        $nY = count($yValues);
-        $nX = count($xValues);
+        $yValueCount = count($yValues);
+        $xValueCount = count($xValues);
 
         //    Define X Values if necessary
-        if ($nX == 0) {
-            $xValues = range(1, $nY);
-        } elseif ($nY != $nX) {
+        if ($xValueCount === 0) {
+            $xValues = range(1, $yValueCount);
+        } elseif ($yValueCount !== $xValueCount) {
             //    Ensure both arrays of points are the same size
             $this->error = true;
         }
 
-        $this->valueCount = $nY;
+        $this->valueCount = $yValueCount;
         $this->xValues = $xValues;
         $this->yValues = $yValues;
     }

diff --git a/src/PhpSpreadsheet/Shared/Trend/ExponentialBestFit.php b/src/PhpSpreadsheet/Shared/Trend/ExponentialBestFit.php
@@ -88,20 +88,17 @@ public function getIntersect($dp = 0)
      *
      * @param float[] $yValues The set of Y-values for this regression
      * @param float[] $xValues The set of X-values for this regression
-     * @param bool $const
      */
-    private function exponentialRegression($yValues, $xValues, $const): void
+    private function exponentialRegression(array $yValues, array $xValues, bool $const): void
     {
-        foreach ($yValues as &$value) {
-            if ($value < 0.0) {
-                $value = 0 - log(abs($value));
-            } elseif ($value > 0.0) {
-                $value = log($value);
-            }
-        }
-        unset($value);
+        $adjustedYValues = array_map(
+            function ($value) {
+                return ($value < 0.0) ? 0 - log(abs($value)) : log($value);
+            },
+            $yValues
+        );
 
-        $this->leastSquareFit($yValues, $xValues, $const);
+        $this->leastSquareFit($adjustedYValues, $xValues, $const);
     }
 
     /**
@@ -116,7 +113,7 @@ public function __construct($yValues, $xValues = [], $const = true)
         parent::__construct($yValues, $xValues);
 
         if (!$this->error) {
-            $this->exponentialRegression($yValues, $xValues, $const);
+            $this->exponentialRegression($yValues, $xValues, (bool) $const);
         }
     }
 }
diff --git a/src/PhpSpreadsheet/Shared/Trend/LinearBestFit.php b/src/PhpSpreadsheet/Shared/Trend/LinearBestFit.php
@@ -56,9 +56,8 @@ public function getEquation($dp = 0)
      *
      * @param float[] $yValues The set of Y-values for this regression
      * @param float[] $xValues The set of X-values for this regression
-     * @param bool $const
      */
-    private function linearRegression($yValues, $xValues, $const): void
+    private function linearRegression(array $yValues, array $xValues, bool $const): void
     {
         $this->leastSquareFit($yValues, $xValues, $const);
     }
@@ -75,7 +74,7 @@ public function __construct($yValues, $xValues = [], $const = true)
         parent::__construct($yValues, $xValues);
 
         if (!$this->error) {
-            $this->linearRegression($yValues, $xValues, $const);
+            $this->linearRegression($yValues, $xValues, (bool) $const);
         }
     }
 }
diff --git a/src/PhpSpreadsheet/Shared/Trend/LogarithmicBestFit.php b/src/PhpSpreadsheet/Shared/Trend/LogarithmicBestFit.php
@@ -48,28 +48,25 @@ public function getEquation($dp = 0)
         $slope = $this->getSlope($dp);
         $intersect = $this->getIntersect($dp);
 
-        return 'Y = ' . $intersect . ' + ' . $slope . ' * log(X)';
+        return 'Y = ' . $slope . ' * log(' . $intersect . ' * X)';
     }
 
     /**
      * Execute the regression and calculate the goodness of fit for a set of X and Y data values.
      *
      * @param float[] $yValues The set of Y-values for this regression
      * @param float[] $xValues The set of X-values for this regression
-     * @param bool $const
      */
-    private function logarithmicRegression($yValues, $xValues, $const): void
+    private function logarithmicRegression(array $yValues, array $xValues, bool $const): void
     {
-        foreach ($xValues as &$value) {
-            if ($value < 0.0) {
-                $value = 0 - log(abs($value));
-            } elseif ($value > 0.0) {
-                $value = log($value);
-            }
-        }
-        unset($value);
+        $adjustedYValues = array_map(
+            function ($value) {
+                return ($value < 0.0) ? 0 - log(abs($value)) : log($value);
+            },
+            $yValues
+        );
 
-        $this->leastSquareFit($yValues, $xValues, $const);
+        $this->leastSquareFit($adjustedYValues, $xValues, $const);
     }
 
     /**
@@ -84,7 +81,7 @@ public function __construct($yValues, $xValues = [], $const = true)
         parent::__construct($yValues, $xValues);
 
         if (!$this->error) {
-            $this->logarithmicRegression($yValues, $xValues, $const);
+            $this->logarithmicRegression($yValues, $xValues, (bool) $const);
         }
     }
 }
diff --git a/src/PhpSpreadsheet/Shared/Trend/PolynomialBestFit.php b/src/PhpSpreadsheet/Shared/Trend/PolynomialBestFit.php
@@ -178,9 +178,8 @@ private function polynomialRegression($order, $yValues, $xValues): void
      * @param int $order Order of Polynomial for this regression
      * @param float[] $yValues The set of Y-values for this regression
      * @param float[] $xValues The set of X-values for this regression
-     * @param bool $const
      */
-    public function __construct($order, $yValues, $xValues = [], $const = true)
+    public function __construct($order, $yValues, $xValues = [])
     {
         parent::__construct($yValues, $xValues);
 

diff --git a/src/PhpSpreadsheet/Shared/Trend/PowerBestFit.php b/src/PhpSpreadsheet/Shared/Trend/PowerBestFit.php
@@ -72,28 +72,23 @@ public function getIntersect($dp = 0)
      *
      * @param float[] $yValues The set of Y-values for this regression
      * @param float[] $xValues The set of X-values for this regression
-     * @param bool $const
      */
-    private function powerRegression($yValues, $xValues, $const): void
+    private function powerRegression(array $yValues, array $xValues, bool $const): void
     {
-        foreach ($xValues as &$value) {
-            if ($value < 0.0) {
-                $value = 0 - log(abs($value));
-            } elseif ($value > 0.0) {
-                $value = log($value);
-            }
-        }
-        unset($value);
-        foreach ($yValues as &$value) {
-            if ($value < 0.0) {
-                $value = 0 - log(abs($value));
-            } elseif ($value > 0.0) {
-                $value = log($value);
-            }
-        }
-        unset($value);
+        $adjustedYValues = array_map(
+            function ($value) {
+                return ($value < 0.0) ? 0 - log(abs($value)) : log($value);
+            },
+            $yValues
+        );
+        $adjustedXValues = array_map(
+            function ($value) {
+                return ($value < 0.0) ? 0 - log(abs($value)) : log($value);
+            },
+            $xValues
+        );
 
-        $this->leastSquareFit($yValues, $xValues, $const);
+        $this->leastSquareFit($adjustedYValues, $adjustedXValues, $const);
     }
 
     /**
@@ -108,7 +103,7 @@ public function __construct($yValues, $xValues = [], $const = true)
         parent::__construct($yValues, $xValues);
 
         if (!$this->error) {
-            $this->powerRegression($yValues, $xValues, $const);
+            $this->powerRegression($yValues, $xValues, (bool) $const);
         }
     }
 }
diff --git a/src/PhpSpreadsheet/Shared/Trend/Trend.php b/src/PhpSpreadsheet/Shared/Trend/Trend.php
@@ -55,10 +55,9 @@ public static function calculate($trendType = self::TREND_BEST_FIT, $yValues = [
         $nX = count($xValues);
 
         //    Define X Values if necessary
-        if ($nX == 0) {
+        if ($nX === 0) {
             $xValues = range(1, $nY);
-            $nX = $nY;
-        } elseif ($nY != $nX) {
+        } elseif ($nY !== $nX) {
             //    Ensure both arrays of points are the same size
             trigger_error('Trend(): Number of elements in coordinate arrays do not match.', E_USER_ERROR);
         }
@@ -84,7 +83,7 @@ public static function calculate($trendType = self::TREND_BEST_FIT, $yValues = [
             case self::TREND_POLYNOMIAL_6:
                 if (!isset(self::$trendCache[$key])) {
                     $order = substr($trendType, -1);
-                    self::$trendCache[$key] = new PolynomialBestFit($order, $yValues, $xValues, $const);
+                    self::$trendCache[$key] = new PolynomialBestFit($order, $yValues, $xValues);
                 }
 
                 return self::$trendCache[$key];
@@ -100,7 +99,7 @@ public static function calculate($trendType = self::TREND_BEST_FIT, $yValues = [
                 if ($trendType != self::TREND_BEST_FIT_NO_POLY) {
                     foreach (self::$trendTypePolynomialOrders as $trendMethod) {
                         $order = substr($trendMethod, -1);
-                        $bestFit[$trendMethod] = new PolynomialBestFit($order, $yValues, $xValues, $const);
+                        $bestFit[$trendMethod] = new PolynomialBestFit($order, $yValues, $xValues);
                         if ($bestFit[$trendMethod]->getError()) {
                             unset($bestFit[$trendMethod]);
                         } else {

diff --git a/tests/PhpSpreadsheetTests/Calculation/Functions/Statistical/LogEstTest.php b/tests/PhpSpreadsheetTests/Calculation/Functions/Statistical/LogEstTest.php
@@ -19,7 +19,7 @@ class LogEstTest extends TestCase
     public function testLOGEST($expectedResult, $yValues, $xValues, $const, $stats): void
     {
         $result = Statistical::LOGEST($yValues, $xValues, $const, $stats);
-
+        //var_dump($result);
         $elements = count($expectedResult);
         for ($element = 0; $element < $elements; ++$element) {
             self::assertEqualsWithDelta($expectedResult[$element], $result[$element], 1E-12);

diff --git a/tests/PhpSpreadsheetTests/Shared/Trend/ExponentialBestFitTest.php b/tests/PhpSpreadsheetTests/Shared/Trend/ExponentialBestFitTest.php
@@ -0,0 +1,49 @@
+<?php
+
+namespace PhpOffice\PhpSpreadsheetTests\Shared\Trend;
+
+use PhpOffice\PhpSpreadsheet\Shared\Trend\ExponentialBestFit;
+use PHPUnit\Framework\TestCase;
+
+class ExponentialBestFitTest extends TestCase
+{
+    /**
+     * @dataProvider providerExponentialBestFit
+     *
+     * @param mixed $expectedSlope
+     * @param mixed $expectedIntersect
+     * @param mixed $expectedGoodnessOfFit
+     * @param mixed $yValues
+     * @param mixed $xValues
+     * @param mixed $expectedEquation
+     */
+    public function testExponentialBestFit(
+        $expectedSlope,
+        $expectedIntersect,
+        $expectedGoodnessOfFit,
+        $expectedEquation,
+        $yValues,
+        $xValues
+    ): void {
+        $bestFit = new ExponentialBestFit($yValues, $xValues);
+        $slope = $bestFit->getSlope(1);
+        self::assertEquals($expectedSlope[0], $slope);
+        $slope = $bestFit->getSlope();
+        self::assertEquals($expectedSlope[1], $slope);
+        $intersect = $bestFit->getIntersect(1);
+        self::assertEquals($expectedIntersect[0], $intersect);
+        $intersect = $bestFit->getIntersect();
+        self::assertEquals($expectedIntersect[1], $intersect);
+
+        $equation = $bestFit->getEquation(2);
+        self::assertEquals($expectedEquation, $equation);
+
+        self::assertSame($expectedGoodnessOfFit[0], $bestFit->getGoodnessOfFit(6));
+        self::assertSame($expectedGoodnessOfFit[1], $bestFit->getGoodnessOfFit());
+    }
+
+    public function providerExponentialBestFit()
+    {
+        return require 'tests/data/Shared/Trend/ExponentialBestFit.php';
+    }
+}