/*
* $RCSfile: NumericToInterpolatedDecDistribution.java,v $
* $Revision: 1.7 $
* $Date: 2007/06/30 17:30:33 $
* $Author: wojna $
*
* Copyright (C) 2002 - 2007 Logic Group, Institute of Mathematics, Warsaw University
*
* This file is part of Rseslib.
*
* Rseslib is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 3 of the License, or
* (at your option) any later version.
*
* Rseslib 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 General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
package rseslib.structure.function.decvector;
import rseslib.structure.function.doubleval.AttributeDoubleFunction;
import rseslib.structure.table.DecisionDistributionsForDiscretizedNumerical;
import rseslib.structure.table.DoubleDataTable;
import rseslib.structure.table.IntervalWithDecisionDistribution;
import rseslib.structure.vector.Vector;
/**
* The function translating the value of a numeric attribute
* into decision distribution vector for this value.
* If the flag m_bConvertToVectDecDifference is off,
* the function returns the real index of the interval
* containing the attribute value. The real index corresponds
* to the interpolation between the interval preceding
* the attribute value and the interval succeeding
* the attribute value.
* If the flag m_bConvertToVectDecDifference is on,
* it implies that the decision attribute has two values
* and a decision vector has two coordinates
* and the function returns the difference between
* the first and the second coordinates
* of the interpolated decision vector.
*
* @author Arkadiusz Wojna
*/
public class NumericToInterpolatedDecDistribution extends AttributeDoubleFunction
{
/** Serialization version. */
private static final long serialVersionUID = 1L;
/**
* Array of successive intervals with decision vectors,
* position 0 contains decision distribution for the null value.
*/
private IntervalWithDecisionDistribution[] m_DecDistributions = null;
/** The minimal value of the attribute in a training set. */
private double m_nMinValue;
/** Length of a single interval obtained from discretisation. */
private double m_nIntervalLength;
/**
* The flag is set to true if the decision attribute has two values
* and then the function returns the difference between
* the first and the second coordinates of the decision vector.
*/
private boolean m_bConvertToVectDecDifference = false;
/**
* Constructor extracts decision vectors
* for succesive value ranges.
* The value ranges are obtained by dividing
* the range of all values into equal intervals.
*
* @param objects Data set that decision vectors for succesive value ranges
* are calculated from.
* @param attrIndex Attribute index.
*/
public NumericToInterpolatedDecDistribution(DoubleDataTable objects, int attrIndex)
{
super(attrIndex);
int noOfIntervals = objects.attributes().nominalDecisionAttribute().noOfValues();
if (noOfIntervals < 5) noOfIntervals = 5;
DecisionDistributionsForDiscretizedNumerical decProbabilities = new DecisionDistributionsForDiscretizedNumerical(objects, attrIndex, noOfIntervals);
m_DecDistributions = decProbabilities.getIntervals();
m_nIntervalLength = decProbabilities.getIntervalLength();
m_nMinValue = objects.getNumericalStatistics(attrIndex).getMinimum();
if (objects.attributes().nominalDecisionAttribute().noOfValues()==2)
m_bConvertToVectDecDifference = true;
else m_bConvertToVectDecDifference = false;
}
/**
* Returns decision vectors for succesive intervals.
*
* @return Decision vectors for succesive intervals.
*/
public Vector[] getDiscretisedDecVectors()
{
Vector[] valueDict = new Vector[m_DecDistributions.length];
for (int val = 0; val < valueDict.length; val++)
valueDict[val] = m_DecDistributions[val].getDecDistribution();
return valueDict;
}
/**
* Returns the value of this function for a given attribute value.
*
* @param attrVal Attribute value.
* @return Value of this function for a given attribute value.
*/
public double doubleVal(double attrVal)
{
if (Double.isNaN(attrVal)) return Double.NaN;
double interval = Double.NaN;
if (m_DecDistributions.length > 1)
interval = (attrVal - m_nMinValue) / m_nIntervalLength + 0.5;
if (m_bConvertToVectDecDifference)
{
if (Double.isNaN(interval)) interval = 0.0;
else if (interval <= 0.0 || interval >= (double)m_DecDistributions.length) return 0.0;
int left = (int)interval;
if (interval == (double)left)
return (m_DecDistributions[left].getDecDistribution().get(0)-m_DecDistributions[left].getDecDistribution().get(1));
double leftWeight = (double)(left + 1) - interval;
double rightWeight = interval - (double)left;
if (interval < 1.0)
return rightWeight*(m_DecDistributions[left+1].getDecDistribution().get(0)-m_DecDistributions[left+1].getDecDistribution().get(1));
if (interval > (double)(m_DecDistributions.length - 1))
return leftWeight*(m_DecDistributions[left].getDecDistribution().get(0)-m_DecDistributions[left].getDecDistribution().get(1));
return (m_DecDistributions[left].getDecDistribution().get(0)*leftWeight+m_DecDistributions[left+1].getDecDistribution().get(0)*rightWeight
-m_DecDistributions[left].getDecDistribution().get(1)*leftWeight-m_DecDistributions[left+1].getDecDistribution().get(1)*rightWeight);
}
if (Double.isNaN(interval)) interval = 0.0;
return interval;
}
}
