/*
 *  RapidMiner
 *
 *  Copyright (C) 2001-2014 by RapidMiner and the contributors
 *
 *  Complete list of developers available at our web site:
 *
 *       http://rapidminer.com
 *
 *  This program is free software: you can redistribute it and/or modify
 *  it under the terms of the GNU Affero General Public License as published by
 *  the Free Software Foundation, either version 3 of the License, or
 *  (at your option) any later version.
 *
 *  This program is distributed in the hope that it will be useful,
 *  but WITHOUT ANY WARRANTY; without even the implied warranty of
 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 *  GNU Affero General Public License for more details.
 *
 *  You should have received a copy of the GNU Affero General Public License
 *  along with this program.  If not, see http://www.gnu.org/licenses/.
 */
package com.rapidminer.operator.performance;

import java.util.Iterator;

import com.rapidminer.example.Attribute;
import com.rapidminer.example.Example;
import com.rapidminer.example.ExampleSet;
import com.rapidminer.operator.OperatorException;
import com.rapidminer.operator.UserError;
import com.rapidminer.tools.math.Averagable;


/**
 * Relative squared error is the total squared error made relative to what the
 * error would have been if the prediction had been the average of the absolute
 * value. As done with the root mean-squared error, the square root of the
 * relative squared error is taken to give it the same dimensions as the
 * predicted values themselves. Also, just like root mean-squared error, this
 * exaggerates the cases in which the prediction error was significantly greater
 * than the mean error.
 * 
 * @author Ingo Mierswa
 *          ingomierswa Exp $
 */
public class RootRelativeSquaredError extends MeasuredPerformance {

	private static final long serialVersionUID = 7781104825149866444L;

	private Attribute predictedAttribute;

	private Attribute labelAttribute;
	
	private Attribute weightAttribute;

	private double deviationSum = 0.0d;

	private double relativeSum = 0.0d;

	private double trueLabelSum = 0.0d;

	private double exampleCounter = 0;

	public RootRelativeSquaredError() {
	}

	public RootRelativeSquaredError(RootRelativeSquaredError rse) {
		super(rse);
		this.deviationSum = rse.deviationSum;
		this.relativeSum = rse.relativeSum;
		this.trueLabelSum = rse.trueLabelSum;
		this.exampleCounter = rse.exampleCounter;
        this.labelAttribute = (Attribute)rse.labelAttribute.clone();
        this.predictedAttribute = (Attribute)rse.predictedAttribute.clone();
        if (rse.weightAttribute != null)
        	this.weightAttribute = (Attribute)rse.weightAttribute.clone();
	}

	@Override
	public String getName() {
		return "root_relative_squared_error";
	}

	@Override
	public String getDescription() {
		return "Averaged root-relative-squared error";
	}

	@Override
	public double getExampleCount() {
		return exampleCounter;
	}

	@Override
	public void startCounting(ExampleSet exampleSet, boolean useExampleWeights) throws OperatorException {
		super.startCounting(exampleSet, useExampleWeights);
		if (exampleSet.size() <= 1)
			throw new UserError(null, 919, getName(), "root relative squared error can only be calculated for test sets with more than 2 examples.");
		this.predictedAttribute = exampleSet.getAttributes().getPredictedLabel();
		this.labelAttribute = exampleSet.getAttributes().getLabel();
		if (useExampleWeights)
			this.weightAttribute = exampleSet.getAttributes().getWeight();
		
		this.trueLabelSum = 0.0d;
		this.deviationSum = 0.0d;
		this.relativeSum = 0.0d;
		this.exampleCounter = 0.0d;
		Iterator<Example> reader = exampleSet.iterator();
		while (reader.hasNext()) {
			Example example = reader.next();
			double label = example.getValue(labelAttribute);
			if (!Double.isNaN(label)) {
				exampleCounter += 1;
				trueLabelSum += label;
			}
		}
	}

	/** Calculates the error for the current example. */
	@Override
	public void countExample(Example example) {
		double plabel;
		double label = example.getValue(labelAttribute);

		if (!predictedAttribute.isNominal()) {
			plabel = example.getValue(predictedAttribute);
		} else {
			String labelS = example.getNominalValue(labelAttribute);
			plabel = example.getConfidence(labelS);
			label = 1.0d;
		}

		double weight = 1.0d;
		if (weightAttribute != null)
			weight = example.getValue(weightAttribute);
		
		double diff = Math.abs(label - plabel);
		deviationSum += diff * diff * weight * weight;
		double relDiff = Math.abs(label - (trueLabelSum / exampleCounter));
		relativeSum += relDiff * relDiff * weight * weight;
	}

	@Override
	public double getMikroAverage() {
		return Math.sqrt(deviationSum / relativeSum);
	}

	@Override
	public double getMikroVariance() {
		return Double.NaN;
	}

	@Override
	public double getFitness() {
		return (-1) * getAverage();
	}

	@Override
	public void buildSingleAverage(Averagable performance) {
		RootRelativeSquaredError other = (RootRelativeSquaredError) performance;
		this.deviationSum += other.deviationSum;
		this.relativeSum += other.relativeSum;
	}
}