Added model

Author: mikoweb

.gitignore

@@ -3,6 +3,7 @@ composer.phar
 /resources/valid.csv
 /resources/training.csv
 /resources/test.csv
+/resources/model.rbx
 
 ###> symfony/framework-bundle ###
 /.env.local

src/Application/ML/LearnerFactory.php

@@ -0,0 +1,32 @@
+<?php
+
+namespace App\Application\ML;
+
+use Rubix\ML\Learner;
+use Rubix\ML\NeuralNet\ActivationFunctions\ReLU;
+use Rubix\ML\NeuralNet\Layers\Activation;
+use Rubix\ML\NeuralNet\Layers\Dense;
+use Rubix\ML\Pipeline;
+use Rubix\ML\Regressors\MLPRegressor;
+use Rubix\ML\Transformers\NumericStringConverter;
+use Rubix\ML\Transformers\ZScaleStandardizer;
+
+/**
+ * @see https://docs.rubixml.com/latest/transformers/z-scale-standardizer.html
+ * @see https://docs.rubixml.com/latest/regressors/mlp-regressor.html
+ */
+class LearnerFactory
+{
+    public static function createLearner(): Learner
+    {
+        return new Pipeline([
+            new NumericStringConverter(),
+            new ZScaleStandardizer(),
+        ], new MLPRegressor([
+            new Dense(8),
+            new Activation(new ReLU()),
+            new Dense(8),
+            new Activation(new ReLU()),
+        ]));
+    }
+}

src/Application/ML/PowerStationDatasetSplitter.php

@@ -8,7 +8,7 @@
 use League\Csv\Exception;
 use League\Csv\UnavailableStream;
 use League\Csv\Writer;
-use Rubix\ML\Datasets\Labeled;
+use Rubix\ML\Datasets\Unlabeled;
 use Rubix\ML\Extractors\CSV;
 use Rubix\ML\Transformers\NumericStringConverter;
 
@@ -26,7 +26,7 @@ public function __construct(
      */
     public function split(): void
     {
-        $dataset = Labeled::fromIterator(new CSV(
+        $dataset = Unlabeled::fromIterator(new CSV(
             $this->appPathResolver->getResourcesPath(FileNames::VALID_DATA),
             header: true,
         ))->apply(new NumericStringConverter());

src/Application/ML/PowerStationReport.php

@@ -0,0 +1,45 @@
+<?php
+
+namespace App\Application\ML;
+
+use App\Application\Path\AppPathResolver;
+use App\Domain\FileNames;
+use Rubix\ML\CrossValidation\Reports\AggregateReport;
+use Rubix\ML\CrossValidation\Reports\ErrorAnalysis;
+use Rubix\ML\Datasets\Labeled;
+use Rubix\ML\Extractors\CSV;
+use Rubix\ML\PersistentModel;
+use Rubix\ML\Persisters\Filesystem;
+use Rubix\ML\Report;
+
+/**
+ * @see https://docs.rubixml.com/latest/cross-validation.html
+ * @see https://docs.rubixml.com/latest/cross-validation/reports/error-analysis.html
+ */
+readonly class PowerStationReport
+{
+    public function __construct(
+        private AppPathResolver $appPathResolver,
+    ) {
+    }
+
+    public function generateReport(): Report
+    {
+        $dataset = Labeled::fromIterator(new CSV(
+            $this->appPathResolver->getResourcesPath(FileNames::TEST_SET),
+            header: false,
+        ))->transformLabels(fn ($value) => (float) $value);
+
+        $estimator = PersistentModel::load(new Filesystem(
+            $this->appPathResolver->getResourcesPath(FileNames::MODEL_FILENAME),
+        ));
+
+        $predictions = $estimator->predict($dataset);
+
+        $report = new AggregateReport([
+            new ErrorAnalysis(),
+        ]);
+
+        return $report->generate($predictions, $dataset->labels());
+    }
+}

src/Application/ML/PowerStationTester.php

@@ -0,0 +1,35 @@
+<?php
+
+namespace App\Application\ML;
+
+use App\Application\Path\AppPathResolver;
+use App\Domain\FileNames;
+use Rubix\ML\CrossValidation\KFold;
+use Rubix\ML\CrossValidation\Metrics\MeanSquaredError;
+use Rubix\ML\Datasets\Labeled;
+use Rubix\ML\Extractors\CSV;
+
+/**
+ * @see https://docs.rubixml.com/latest/cross-validation.html#validators
+ * @see https://docs.rubixml.com/latest/cross-validation/metrics/mean-squared-error.html
+ */
+readonly class PowerStationTester
+{
+    public function __construct(
+        private AppPathResolver $appPathResolver,
+    ) {
+    }
+
+    public function test(int $foldsNumber = 5): float
+    {
+        $dataset = Labeled::fromIterator(new CSV(
+            $this->appPathResolver->getResourcesPath(FileNames::VALID_DATA),
+            header: true,
+        ))->transformLabels(fn ($value) => (float) $value);
+
+        $estimator = LearnerFactory::createLearner();
+        $validator = new KFold($foldsNumber);
+
+        return $validator->test($estimator, $dataset, new MeanSquaredError());
+    }
+}

src/Application/ML/PowerStationTrainer.php

@@ -0,0 +1,34 @@
+<?php
+
+namespace App\Application\ML;
+
+use App\Application\Path\AppPathResolver;
+use App\Domain\FileNames;
+use Rubix\ML\Datasets\Labeled;
+use Rubix\ML\Extractors\CSV;
+use Rubix\ML\PersistentModel;
+use Rubix\ML\Persisters\Filesystem;
+
+readonly class PowerStationTrainer
+{
+    public function __construct(
+        private AppPathResolver $appPathResolver,
+    ) {
+    }
+
+    public function train(bool $history = false): void
+    {
+        $dataset = Labeled::fromIterator(new CSV(
+            $this->appPathResolver->getResourcesPath(FileNames::TRAINING_SET),
+            header: false,
+        ))->transformLabels(fn ($value) => (float) $value);
+
+        $estimator = new PersistentModel(
+            LearnerFactory::createLearner(),
+            new Filesystem($this->appPathResolver->getResourcesPath(FileNames::MODEL_FILENAME), $history),
+        );
+
+        $estimator->train($dataset);
+        $estimator->save();
+    }
+}

src/Domain/FileNames.php

@@ -8,4 +8,5 @@ final class FileNames
     public const string VALID_DATA = 'valid.csv';
     public const string TRAINING_SET = 'training.csv';
     public const string TEST_SET = 'test.csv';
+    public const string MODEL_FILENAME = 'model.rbx';
 }

src/UI/CLI/SolveTaskCommand.php

@@ -4,6 +4,9 @@
 
 use App\Application\Analytics\PowerCorrelationAnalyzer;
 use App\Application\ML\PowerStationDatasetSplitter;
+use App\Application\ML\PowerStationReport;
+use App\Application\ML\PowerStationTester;
+use App\Application\ML\PowerStationTrainer;
 use App\Domain\PowerStation;
 use App\Infrastructure\Reader\PowerStationDatasetReader;
 use App\Infrastructure\Writer\PowerStationValidDataWriter;
@@ -26,6 +29,9 @@ public function __construct(
         private readonly ValidatorInterface $validator,
         private readonly PowerStationValidDataWriter $powerStationValidDataWriter,
         private readonly PowerStationDatasetSplitter $powerStationDatasetSplitter,
+        private readonly PowerStationTrainer $powerStationTrainer,
+        private readonly PowerStationTester $powerStationTester,
+        private readonly PowerStationReport $powerStationReport,
     ) {
         parent::__construct();
     }
@@ -85,6 +91,16 @@ protected function execute(InputInterface $input, OutputInterface $output): int
         $io->info('Splitting the data into training and test sets...');
         $this->powerStationDatasetSplitter->split();
 
+        $io->info('Training the model...');
+        $this->powerStationTrainer->train();
+
+        $io->info('Testing the model...');
+        $score = $this->powerStationTester->test();
+        $io->success(sprintf('MeanSquaredError: %f', abs($score)));
+
+        $io->info('Generating model report...');
+        dump($this->powerStationReport->generateReport());
+
         $io->success('OK');
 
         return Command::SUCCESS;