moa.classifiers.rules.multilabel.attributeclassobservers.MultiLabelBSTree Maven / Gradle / Ivy

Go to download

Show more of this group Show more artifacts with this name
Show all versions of moa Show documentation

Massive On-line Analysis is an environment for massive data mining. MOA provides a framework for data stream mining and includes tools for evaluation and a collection of machine learning algorithms. Related to the WEKA project, also written in Java, while scaling to more demanding problems.

The newest version!

/*
 *    MultiLabelBSTree.java
 *    Copyright (C) 2017 University of Porto, Portugal
 *    @author J. Duarte, J. Gama
 *
 *    Licensed under the Apache License, Version 2.0 (the "License");
 *    you may not use this file except in compliance with the License.
 *    You may obtain a copy of the License at
 *
 *      http://www.apache.org/licenses/LICENSE-2.0
 *
 *    Unless required by applicable law or agreed to in writing, software
 *    distributed under the License is distributed on an "AS IS" BASIS,
 *    WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 *    See the License for the specific language governing permissions and
 *    limitations under the License.
 *
 *
 */

package moa.classifiers.rules.multilabel.attributeclassobservers;

import java.io.Serializable;

import moa.classifiers.rules.core.NumericRulePredicate;
import moa.classifiers.rules.core.Utils;
import moa.classifiers.rules.multilabel.core.AttributeExpansionSuggestion;
import moa.classifiers.rules.multilabel.core.splitcriteria.MultiLabelSplitCriterion;
import moa.core.DoubleVector;
import moa.core.ObjectRepository;
import moa.options.AbstractOptionHandler;
import moa.tasks.TaskMonitor;

import com.github.javacliparser.IntOption;

/**
 * Binary search tree for AMRules splitting points determination
 */



public class MultiLabelBSTree extends AbstractOptionHandler implements NumericStatisticsObserver {

	/**
	 * 
	 */
	public IntOption maxNodesOption = new IntOption("maxNodesOption", 'z', "Maximum number of nodes", 50, 0, Integer.MAX_VALUE);
	protected int maxNodes;
	protected int numNodes;

	private static final long serialVersionUID = 1L;

	protected Node root = null;

	protected DoubleVector [] leftStatistics;
	protected DoubleVector [] rightStatistics;

	@Override
	public void observeAttribute(double inputAttributeValue,
			DoubleVector[] statistics) {
		if (!Double.isNaN(inputAttributeValue))
		{
			if (this.root == null) {
				this.root = new Node(inputAttributeValue, statistics);
				maxNodes=maxNodesOption.getValue();
			} else {
				this.root.observeAttribute(inputAttributeValue, statistics);
			}
		}

	}


	@Override
	public AttributeExpansionSuggestion getBestEvaluatedSplitSuggestion(
			MultiLabelSplitCriterion criterion, DoubleVector[] preSplitStatistics, int inputAttributeIndex) {
		// Initialize global variables
		int numOutputs=preSplitStatistics.length;
		leftStatistics=new DoubleVector[numOutputs];
		rightStatistics=new DoubleVector[numOutputs];
		for (int i=0; i< numOutputs; i++)
		{
			leftStatistics[i]=new DoubleVector(new double [preSplitStatistics[i].numValues()]); //sets statistics to zeros
			rightStatistics[i]=new DoubleVector(preSplitStatistics[i]);
		}
		AttributeExpansionSuggestion ret=searchForBestSplitOption(this.root, null, criterion, preSplitStatistics, inputAttributeIndex);
		leftStatistics=null;
		rightStatistics=null;
		return ret;
	}

	protected AttributeExpansionSuggestion searchForBestSplitOption(Node currentNode, AttributeExpansionSuggestion currentBestOption, MultiLabelSplitCriterion criterion, DoubleVector [] preSplitStatistics, int inputAttributeIndex) {
		// Return null if the current node is null or we have finished looking through all the possible splits
		if (currentNode == null) { // TODO: JD check || countRightTotal == 0.0
			return currentBestOption;
		}

		if (currentNode.left != null) {
			currentBestOption = searchForBestSplitOption(currentNode.left, currentBestOption, criterion, preSplitStatistics, inputAttributeIndex);
		}
		for (int i=0; i currentBestOption.merit)) {
			currentBestOption= new AttributeExpansionSuggestion(new NumericRulePredicate(inputAttributeIndex, currentNode.cutPoint, true), Utils.copy(postSplitDists), merit);
		}

		if (currentNode.right != null) {
			currentBestOption = searchForBestSplitOption(currentNode.right, currentBestOption, criterion, preSplitStatistics, inputAttributeIndex);
		}
		for (int i=0; i