phpcs fixes and classifier bug fixes/searching

Algorithm/Sort/CycleSort.php

@@ -29,14 +29,13 @@ class CycleSort implements SortInterface
      */
     public static function sort(array $list, int $order = SortOrder::ASC) : array
     {
-        $writes = 0;
+        $n = \count($list);
-        $n      = \count($list);
 
         if ($n < 2) {
             return $list;
         }
 
-        for ($start = 0; $start < \count($list) - 1; ++$start) {
+        for ($start = 0; $start < $n - 1; ++$start) {
             $item = $list[$start];
 
             $pos     = $start;
@@ -58,7 +57,6 @@ class CycleSort implements SortInterface
             $old        = $list[$pos];
             $list[$pos] = $item;
             $item       = $old;
-            ++$writes;
 
             while ($pos !== $start) {
                 $pos     = $start;
@@ -76,7 +74,6 @@ class CycleSort implements SortInterface
                 $old        = $list[$pos];
                 $list[$pos] = $item;
                 $item       = $old;
-                ++$writes;
             }
         }
 

Algorithm/Sort/StoogeSort.php

@@ -36,7 +36,7 @@ class StoogeSort implements SortInterface
         }
 
         $copy = $list;
-        self::stoogeSort($copy, 0, $n - 1, $order);
+        self::stooge($copy, 0, $n - 1, $order);
 
         return $copy;
     }
@@ -53,7 +53,7 @@ class StoogeSort implements SortInterface
      *
      * @since 1.0.0
      */
-    private static function stoogeSort(array &$list, int $lo, int $hi, int $order) : void
+    private static function stooge(array &$list, int $lo, int $hi, int $order) : void
     {
         if ($lo >= $hi) {
             return;
@@ -68,9 +68,9 @@ class StoogeSort implements SortInterface
         if ($hi - $lo + 1 > 2) {
             $t = (int) (($hi - $lo + 1) / 3);
 
-            self::stoogeSort($list, $lo, $hi - $t, $order);
+            self::stooge($list, $lo, $hi - $t, $order);
-            self::stoogeSort($list, $lo + $t, $hi, $order);
+            self::stooge($list, $lo + $t, $hi, $order);
-            self::stoogeSort($list, $lo, $hi - $t, $order);
+            self::stooge($list, $lo, $hi - $t, $order);
         }
     }
 }

Algorithm/Sort/TimSort.php

@@ -50,8 +50,7 @@ class TimSort implements SortInterface
                 $temp = $list[$j];
                 $c    = $j - 1;
 
-                while ($c >= $lo && $list[$c]->compare($temp, $order))
+                while ($c >= $lo && $list[$c]->compare($temp, $order)) {
-                {
                     $list[$c + 1] = $list[$c];
                     --$c;
                 }

Math/Stochastic/NaiveBayesClassifier.php

@@ -24,7 +24,7 @@ use phpOMS\Math\Statistic\MeasureOfDispersion;
  * @link    https://orange-management.org
  * @since   1.0.0
  */
-class NaiveBayesFilter
+class NaiveBayesClassifier
 {
     /**
      * Dictionary of different criterias.
@@ -85,7 +85,7 @@ class NaiveBayesFilter
                 if (!isset($this->probabilities['attr'][$attr])) {
                     $this->probabilities['attr'][$attr] = [
                         'count'    => 0,
-                        'evidence' => 0.0,
+                        'data'     => [],
                     ];
                 }
 
@@ -152,15 +152,19 @@ class NaiveBayesFilter
                     if (isset($this->dict[$criteria][$attr]['data'][$word])
                         && $this->dict[$criteria][$attr]['data'][$word] >= $minimum
                     ) {
-                        $p = $this->dict[$criteria][$attr]['data'][$word] / \array_sum($this->dict[$criteria][$attr]['data'])
+                        $p = ($this->dict[$criteria][$attr]['data'][$word] / \array_sum($this->dict[$criteria][$attr]['data']))
-                            / $this->probabilities['attr'][$attr]['evidence'];
+                            * ($this->probabilities['criteria'][$criteria]['count'] / $this->probabilities['count'])
+                            / $this->probabilities['attr'][$attr]['data'][$word];
 
                         $n += \log(1 - $p) - \log($p);
                     }
                 }
             } else {
+                // todo: add probability of criteria / total?
                 $p = 1 / \sqrt(2 * \M_PI * $this->probabilities['criteria'][$criteria]['attr'][$attr]['variance'])
-                    * \exp(-($value - $this->probabilities['criteria'][$criteria]['attr'][$attr]['mean']) ** 2 / (2 * $this->probabilities['criteria'][$criteria]['attr'][$attr]['variance'] ** 2));
+                    * \exp(-($value - $this->probabilities['criteria'][$criteria]['attr'][$attr]['mean']) / (2 * $this->probabilities['criteria'][$criteria]['attr'][$attr]['variance']));
+
+                var_dump($p);
 
                 $n += \log(1 - $p) - \log($p);
             }
@@ -178,20 +182,34 @@ class NaiveBayesFilter
      */
     private function cache() : void
     {
+        $this->probabilities['attr'] = [];
+
         foreach ($this->dict as $criteria => $subDict) {
             foreach ($subDict as $attr => $valueArray) {
                 if ($valueArray['type'] === 2) {
                     $this->probabilities['criteria'][$criteria]['attr'][$attr]['mean']     = Average::arithmeticMean($this->dict[$criteria][$attr]['data']);
-                    $this->probabilities['criteria'][$criteria]['attr'][$attr]['variance'] = MeasureOfDispersion::empiricalVariance($this->dict[$criteria][$attr]['data']);
+                    $this->probabilities['criteria'][$criteria]['attr'][$attr]['variance'] = MeasureOfDispersion::empiricalVariance($this->dict[$criteria][$attr]['data'], [], $this->probabilities['criteria'][$criteria]['attr'][$attr]['mean']);
+
+                    // \var_dump($criteria);
+                    // \var_dump($attr);
+                    // \var_dump($this->probabilities['criteria'][$criteria]['attr'][$attr]['mean']); // good
+                    // \var_dump($this->probabilities['criteria'][$criteria]['attr'][$attr]['variance']); // bad
                 } else {
-                    $this->probabilities['attr'][$attr]['evidence'] = 0.0;
+                    if (!isset( $this->probabilities['attr'][$attr])) {
+                        $this->probabilities['attr'] = [$attr => ['data' => []]];
+                    }
 
                     foreach ($valueArray['data'] as $word => $count) {
                         if (!isset($this->dict[$criteria][$attr]['data'][$word])) {
                             continue;
                         }
 
-                        $this->probabilities['attr'][$attr]['evidence'] += $this->dict[$criteria][$attr]['data'][$word] / \array_sum($this->dict[$criteria][$attr]['data']);
+                        if (!isset($this->probabilities['attr'][$attr]['data'][$word])) {
+                            $this->probabilities['attr'][$attr]['data'][$word] = 0.0;
+                        }
+
+                        $this->probabilities['attr'][$attr]['data'][$word] += ($this->dict[$criteria][$attr]['data'][$word] / \array_sum($this->dict[$criteria][$attr]['data']))
+                            * ($this->probabilities['criteria'][$criteria]['count'] / $this->probabilities['count']);
                     }
                 }
             }

tests/Math/Stochastic/NaiveBayesClassifierTest.php

@@ -14,12 +14,12 @@
 
 namespace phpOMS\tests\Math\Stochastic;
 
-use phpOMS\Math\Stochastic\NaiveBayesFilter;
+use phpOMS\Math\Stochastic\NaiveBayesClassifier;
 
 /**
  * @internal
  */
-class NaiveBayesFilterTest extends \PHPUnit\Framework\TestCase
+class NaiveBayesClassifierTest extends \PHPUnit\Framework\TestCase
 {
     const PLAY = [
         ['weather' => ['Overcast']],
@@ -55,28 +55,28 @@ class NaiveBayesFilterTest extends \PHPUnit\Framework\TestCase
         ['height' => 5.75, 'weight' => 150, 'foot' => 9],
     ];
 
-    public function testTextFilter() : void
+    public function testTextClassifier() : void
     {
-        $filter = new NaiveBayesFilter();
+        $filter = new NaiveBayesClassifier();
         $filter->train('play', self::PLAY);
         $filter->train('noplay', self::NO_PLAY);
 
         self::assertEqualsWithDelta(
-            0.64,
+            0.6,
             $filter->match('play', ['weather' => ['Sunny']], 1),
             0.01
         );
     }
 
-    public function testNumericFilter() : void
+    public function testNumericClassifier() : void
     {
-        $filter = new NaiveBayesFilter();
+        $filter = new NaiveBayesClassifier();
         $filter->train('male', self::MALE);
         $filter->train('female', self::FEMALE);
 
         self::assertEqualsWithDelta(
             0.64,
-            $filter->match('play', ['height' => 6, 'weight' => 130, 'foot' => 8]),
+            $filter->match('female', ['height' => 6, 'weight' => 130, 'foot' => 8]),
             0.01
         );
     }