package org.jmat.test;

import org.jmat.data.*;
import org.jmat.data.matrixDecompositions.*;
import org.jmat.gui.*;

/**
 * <p>Description : Principal Component Analysis</p>
 * <p>Copyright : GPL</p>
 * @author Yann RICHET
 */

public class PCA {

	private AbstractMatrix covariance;
	private AbstractMatrix EigenVectors;
	private AbstractMatrix EigenValues;

	public PCA(AbstractMatrix X) {
		covariance = new RandomMatrix(X).covariance();
		EigenvalueDecomposition e = covariance.eig();
		EigenVectors = e.getV();
		EigenValues = e.getD();
	}

	public AbstractMatrix getVectors() {
		return EigenVectors;
	}

	public AbstractMatrix getValues() {
		return EigenValues;
	}

	public static void main(String[] arg) {
		//construct the matrix X
		AbstractMatrix x1 = RandomMatrix.normal(100, 1, 0, 1);
		AbstractMatrix x2 = RandomMatrix.normal(100, 1, 0, 1);
		AbstractMatrix X = x1.plus(x2).mergeColumns(x2);

		PCA pca = new PCA(X);

		//display a Frame with data in a 2D-Plot and EigenValues and EigenVectors in the command line.
		new FrameView(X.toPlot2DPanel("[x1+x2,x2]", PlotPanel.SCATTER));
		pca.getValues().toCommandLine("EigenValues");
		pca.getVectors().toCommandLine("EigenVectors");
	}
}