Class Bagging

JSAT/src/jsat/classifiers/boosting/Bagging.java:31–513

An implementation of Bootstrap Aggregating, as described by LEO BREIMAN in "Bagging Predictors". Bagging is an ensemble learner, it takes a weak learner and trains several to create a better over result. Bagging is particularly useful when the base classifier has some amount of predictive

 * @author Edward Raff
 */
public class Bagging implements Classifier, Regressor, Parameterized
{

	private static final long serialVersionUID = -6566453570170428838L;
	private Classifier baseClassifier;
    private Regressor baseRegressor;
    private CategoricalData predicting;
    private int extraSamples;
    private int rounds;
    private boolean simultaniousTraining;
    private Random random;
    private List learners;
    
    /**
     * The number of rounds of bagging that will be used by default in the constructor: {@value #DEFAULT_ROUNDS}
     */
    public static final int DEFAULT_ROUNDS = 20;
    /**
     * The number of extra samples to take when bagging in each round used by default in the constructor: {@value #DEFAULT_EXTRA_SAMPLES}
     */
    public static final int DEFAULT_EXTRA_SAMPLES = 0;
    /**
     * The default behavior for parallel training, as specified by {@link #setSimultaniousTraining(boolean) } is {@value #DEFAULT_SIMULTANIOUS_TRAINING}
     */
    public static final boolean DEFAULT_SIMULTANIOUS_TRAINING = true;
    
    /**
     * Creates a new Bagger for classification. This can not be changed after construction.
     * 
     * @param baseClassifier the base learner to use.
     */
    public Bagging(Classifier baseClassifier)
    {
        this(baseClassifier, DEFAULT_EXTRA_SAMPLES, DEFAULT_SIMULTANIOUS_TRAINING);
    }

    /**
     * Creates a new Bagger for classification. This can not be changed after construction.
     * 
     * @param baseClassifier the base learner to use.
     * @param extraSamples how many extra samples past the training size to take
     * @param simultaniousTraining controls whether base learners are trained sequentially or simultaneously 
     */
    public Bagging(Classifier baseClassifier, int extraSamples, boolean simultaniousTraining)
    {
        this(baseClassifier, extraSamples, simultaniousTraining, DEFAULT_ROUNDS, new Random(1));
    }
    
    /**
     * Creates a new Bagger for classification. This can not be changed after construction.
     * 
     * @param baseClassifier the base learner to use.
     * @param extraSamples how many extra samples past the training size to take
     * @param simultaniousTraining controls whether base learners are trained sequentially or simultaneously 
     * @param rounds how many rounds of bagging to perform. 
     * @param random the source of randomness for sampling 
     */
    public Bagging(Classifier baseClassifier, int extraSamples, boolean simultaniousTraining, int rounds, Random random)