Implement NumberConverter (#377)

Author: akondas

Diff for: src/FeatureExtraction/TfIdfTransformer.php

@@ -30,7 +30,7 @@ public function fit(array $samples, ?array $targets = null): void
         }
     }
 
-    public function transform(array &$samples): void
+    public function transform(array &$samples, ?array &$targets = null): void
     {
         foreach ($samples as &$sample) {
             foreach ($sample as $index => &$feature) {

Diff for: src/FeatureExtraction/TokenCountVectorizer.php

@@ -46,7 +46,7 @@ public function fit(array $samples, ?array $targets = null): void
         $this->buildVocabulary($samples);
     }
 
-    public function transform(array &$samples): void
+    public function transform(array &$samples, ?array &$targets = null): void
     {
         array_walk($samples, function (string &$sample): void {
             $this->transformSample($sample);

Diff for: src/FeatureSelection/SelectKBest.php

@@ -56,7 +56,7 @@ public function fit(array $samples, ?array $targets = null): void
         $this->keepColumns = array_slice($sorted, 0, $this->k, true);
     }
 
-    public function transform(array &$samples): void
+    public function transform(array &$samples, ?array &$targets = null): void
     {
         if ($this->keepColumns === null) {
             return;

Diff for: src/FeatureSelection/VarianceThreshold.php

@@ -48,7 +48,7 @@ public function fit(array $samples, ?array $targets = null): void
         }
     }
 
-    public function transform(array &$samples): void
+    public function transform(array &$samples, ?array &$targets = null): void
     {
         foreach ($samples as &$sample) {
             $sample = array_values(array_intersect_key($sample, $this->keepColumns));

Diff for: src/Preprocessing/Imputer.php

@@ -49,7 +49,7 @@ public function fit(array $samples, ?array $targets = null): void
         $this->samples = $samples;
     }
 
-    public function transform(array &$samples): void
+    public function transform(array &$samples, ?array &$targets = null): void
     {
         if ($this->samples === []) {
             throw new InvalidOperationException('Missing training samples for Imputer.');

Diff for: src/Preprocessing/LabelEncoder.php

@@ -22,7 +22,7 @@ public function fit(array $samples, ?array $targets = null): void
         }
     }
 
-    public function transform(array &$samples): void
+    public function transform(array &$samples, ?array &$targets = null): void
     {
         foreach ($samples as &$sample) {
             $sample = $this->classes[(string) $sample];

Diff for: src/Preprocessing/Normalizer.php

@@ -66,7 +66,7 @@ public function fit(array $samples, ?array $targets = null): void
         $this->fitted = true;
     }
 
-    public function transform(array &$samples): void
+    public function transform(array &$samples, ?array &$targets = null): void
     {
         $methods = [
             self::NORM_L1 => 'normalizeL1',

Diff for: src/Preprocessing/NumberConverter.php

@@ -0,0 +1,47 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Phpml\Preprocessing;
+
+final class NumberConverter implements Preprocessor
+{
+    /**
+     * @var bool
+     */
+    private $transformTargets;
+
+    /**
+     * @var mixed
+     */
+    private $nonNumericPlaceholder;
+
+    /**
+     * @param mixed $nonNumericPlaceholder
+     */
+    public function __construct(bool $transformTargets = false, $nonNumericPlaceholder = null)
+    {
+        $this->transformTargets = $transformTargets;
+        $this->nonNumericPlaceholder = $nonNumericPlaceholder;
+    }
+
+    public function fit(array $samples, ?array $targets = null): void
+    {
+        //nothing to do
+    }
+
+    public function transform(array &$samples, ?array &$targets = null): void
+    {
+        foreach ($samples as &$sample) {
+            foreach ($sample as &$feature) {
+                $feature = is_numeric($feature) ? (float) $feature : $this->nonNumericPlaceholder;
+            }
+        }
+
+        if ($this->transformTargets && is_array($targets)) {
+            foreach ($targets as &$target) {
+                $target = is_numeric($target) ? (float) $target : $this->nonNumericPlaceholder;
+            }
+        }
+    }
+}

Diff for: src/Transformer.php

@@ -11,5 +11,5 @@ interface Transformer
      */
     public function fit(array $samples, ?array $targets = null): void;
 
-    public function transform(array &$samples): void;
+    public function transform(array &$samples, ?array &$targets = null): void;
 }

Diff for: tests/Preprocessing/NumberConverterTest.php

@@ -0,0 +1,47 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Phpml\Tests\Preprocessing;
+
+use Phpml\Preprocessing\NumberConverter;
+use PHPUnit\Framework\TestCase;
+
+final class NumberConverterTest extends TestCase
+{
+    public function testConvertSamples(): void
+    {
+        $samples = [['1', '-4'], ['2.0', 3.0], ['3', '112.5'], ['5', '0.0004']];
+        $targets = ['1', '1', '2', '2'];
+
+        $converter = new NumberConverter();
+        $converter->transform($samples, $targets);
+
+        self::assertEquals([[1.0, -4.0], [2.0, 3.0], [3.0, 112.5], [5.0, 0.0004]], $samples);
+        self::assertEquals(['1', '1', '2', '2'], $targets);
+    }
+
+    public function testConvertTargets(): void
+    {
+        $samples = [['1', '-4'], ['2.0', 3.0], ['3', '112.5'], ['5', '0.0004']];
+        $targets = ['1', '1', '2', 'not'];
+
+        $converter = new NumberConverter(true);
+        $converter->transform($samples, $targets);
+
+        self::assertEquals([[1.0, -4.0], [2.0, 3.0], [3.0, 112.5], [5.0, 0.0004]], $samples);
+        self::assertEquals([1.0, 1.0, 2.0, null], $targets);
+    }
+
+    public function testConvertWithPlaceholder(): void
+    {
+        $samples = [['invalid'], ['13.5']];
+        $targets = ['invalid', '2'];
+
+        $converter = new NumberConverter(true, 'missing');
+        $converter->transform($samples, $targets);
+
+        self::assertEquals([['missing'], [13.5]], $samples);
+        self::assertEquals(['missing', 2.0], $targets);
+    }
+}