Add (unfinished) PCADemo

Author: Stefan Hahmann

pom.xml

@@ -129,6 +129,14 @@
             <version>v2.6.4</version>
         </dependency>
 
+        <!-- PCA -->
+        <!-- Example: https://spark.apache.org/docs/latest/mllib-dimensionality-reduction#principal-component-analysis-pca -->
+        <dependency>
+            <groupId>org.apache.spark</groupId>
+            <artifactId>spark-mllib_2.13</artifactId>
+            <version>3.5.3</version>
+        </dependency>
+
         <!-- Standardization for UMAP preprocessing -->
         <dependency>
             <groupId>org.apache.commons</groupId>

src/test/java/org/mastodon/mamut/feature/dimensionalityreduction/pca/PCADemo.java

@@ -0,0 +1,43 @@
+package org.mastodon.mamut.feature.dimensionalityreduction.pca;
+
+import java.util.ArrayList;
+import java.util.List;
+
+import org.apache.spark.api.java.JavaRDD;
+import org.apache.spark.api.java.JavaSparkContext;
+import org.apache.spark.mllib.linalg.Matrix;
+import org.apache.spark.mllib.linalg.Vector;
+import org.apache.spark.mllib.linalg.Vectors;
+import org.apache.spark.mllib.linalg.distributed.RowMatrix;
+import org.mastodon.mamut.feature.dimensionalityreduction.PlotPoints;
+import org.mastodon.mamut.feature.dimensionalityreduction.RandomDataTools;
+
+public class PCADemo
+{
+	// TODO: check https://stackoverflow.com/questions/10604507/pca-implementation-in-java
+	public static void main( String[] args )
+	{
+		double[][] inputData = RandomDataTools.generateSampleData();
+		double[][] result = setUpPCA( inputData );
+		PlotPoints.plot( inputData, result, resultValues -> resultValues[ 0 ] > 10 );
+	}
+
+	static double[][] setUpPCA( double[][] inputData )
+	{
+		try (JavaSparkContext jsc = new JavaSparkContext( "local", "PCA" ))
+		{
+			List< Vector > data = new ArrayList<>();
+			for ( final double[] row : inputData )
+				data.add( Vectors.dense( row ) );
+			JavaRDD< Vector > rows = jsc.parallelize( data );
+			// Create a RowMatrix from JavaRDD<Vector>.
+			RowMatrix rowMatrix = new RowMatrix( rows.rdd() );
+			// Compute the top 2 principal components.
+			// Principal components are stored in a local dense matrix.
+			Matrix pc = rowMatrix.computePrincipalComponents( 2 );
+			// Project the rows to the linear space spanned by the top 4 principal components.
+			RowMatrix projected = rowMatrix.multiply( pc );
+			return projected.rows().toJavaRDD().collect().stream().map( Vector::toArray ).toArray( double[][]::new );
+		}
+	}
+}