Class RandomForest

JSAT/src/jsat/classifiers/trees/RandomForest.java:45–546

Random Forest is an extension of Bagging that is applied only to DecisionTree DecisionTrees. It works in a similar manner, but also only uses a random sub set of the features for each tree trained. This provides increased performance in accuracy of predictions, and reduced training t

 * @see Bagging
 */
public class RandomForest implements Classifier, Regressor, Parameterized
{
    //TODO implement Out of Bag estimates of proximity, importance, and outlier detection 
    
    private static final long serialVersionUID = 2725020584282958141L;
    /**
     * Only used when training for a classification problem
     */
    private CategoricalData predicting;
    private int extraSamples;
    /**
     * Setting the number of features to use. Default value is -1, indicating the heuristic 
     * of sqrt(N) or N/3 should be used for classification and regression respectively. This
     * value should be set away from -1 before training work begins, and set back if it 
     * was not set explicitly by the used
     */
    private int featureSamples;
    private int maxForestSize;
    private boolean useOutOfBagError = false;
    private boolean useOutOfBagImportance = false;
    private TreeFeatureImportanceInference importanceMeasure = new MDI();
    private OnLineStatistics[] feature_importance = null;
    private double outOfBagError;
    private RandomDecisionTree baseLearner;
    private List<DecisionTree> forest;
    
    public RandomForest()
    {
        this(100);
    }

    public RandomForest(int maxForestSize)
    {
        setExtraSamples(0);
        setMaxForestSize(maxForestSize);
        autoFeatureSample();
        baseLearner = new RandomDecisionTree(1, Integer.MAX_VALUE, 3, TreePruner.PruningMethod.NONE, 1e-15);
        baseLearner.setGainMethod(ImpurityMeasure.GINI);
    }
    
    /**
     * RandomForest performs Bagging. Bagging samples from the training set with replacement, and draws 
     * a sampleWithReplacement at least as large as the training set. This controls how many extra samples are 
     * taken. If negative, fewer samples will be taken. Using negative values is not recommended. 
     * 
     * @param i how many extra samples to take
     */
    public void setExtraSamples(int i)
    {
        extraSamples = i;
    }

    public int getExtraSamples()
    {
        return extraSamples;
    }

    /**