/*
* $RCSfile: TestResult.java,v $
* $Revision: 1.21 $
* $Date: 2007/07/07 15:56:01 $
* $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.processing.classification;
import java.io.*;
import java.util.Enumeration;
import java.util.Properties;
import rseslib.structure.attribute.NominalAttribute;
import rseslib.system.Report;
/**
* Classification accuracy for the whole test set
* and for particular decision classes
* and confusion matrix.
*
* @author Arkadiusz Wojna
*/
public class TestResult implements Serializable
{
/** Serialization version. */
private static final long serialVersionUID = 1L;
/** Dictionary of statistics specific to a classifier. */
private Properties m_Statistics;
/** Decision attribute with value dictionary. */
private NominalAttribute m_DecisionAttribute;
/** Decision distribution in the test set. */
private int[] m_arrDecDistr;
/** Confusion matrix obtained in classification. */
private int[][] m_arrConfusionMatrix;
/** Number of all objects tested. */
private int m_nAll = 0;
/** Number of objects that are assigned with a decision by a classifier. */
private int m_nCovered = 0;
/** Number of correctly classfied objects. */
private int m_nCorrect = 0;
/**
* Constructor.
*
* @param decAttr Information about decision attribute.
* @param decDistr Distribution of objects for locally coded decisions.
* @param confusionMatrix Confusion matrix obtained in classification.
* The i-th row represents the distribution of the decisions assigned
* for objects with the real decision coded locally by i.
* @param statistics Dictionary of general statistics.
*/
public TestResult(NominalAttribute decAttr, int[] decDistr, int[][] confusionMatrix, Properties statistics)
{
if (decDistr.length!=confusionMatrix.length) throw new RuntimeException("The length of the total decision distribution and the confusion matrix dimension do not match");
m_Statistics = statistics;
m_DecisionAttribute = decAttr;
m_arrDecDistr = decDistr;
m_arrConfusionMatrix = confusionMatrix;
for (int dec = 0; dec < m_arrDecDistr.length; dec++)
{
m_nAll += m_arrDecDistr[dec];
m_nCorrect += m_arrConfusionMatrix[dec][dec];
for (int clDec = 0; clDec < m_arrConfusionMatrix[dec].length; clDec++)
m_nCovered += m_arrConfusionMatrix[dec][clDec];
}
}
/**
* Returns the information about the decision attribute.
*
* @return Information about the decision attribute.
*/
public NominalAttribute decisionAttribute()
{
return m_DecisionAttribute;
}
/**
* Returns classification accuracy.
*
* @return Classification accuracy.
*/
public double getAccuracy()
{
return (double)m_nCorrect/(double)m_nAll;
}
/**
* Returns classification accuracy for a given decision class.
*
* @param dec Decision.
* @return Classification accuracy for a given decision class.
*/
public double getDecAccuracy(double dec)
{
int decLocCode = m_DecisionAttribute.localValueCode(dec);
if (m_arrDecDistr[decLocCode]==0) return 0.0;
else return (double)m_arrConfusionMatrix[decLocCode][decLocCode]/(double)m_arrDecDistr[decLocCode];
}
/**
* Returns the number of objects with the real decision realDec
* that were classified with the decision assignedDec.
*
* @param realDec Real decision of the objects to be counted.
* @param assignedDec The decision assigned by the classifier.
* @return Number of objects.
*/
public int getNoOfObjects(double realDec, double assignedDec)
{
return m_arrConfusionMatrix[m_DecisionAttribute.localValueCode(realDec)][m_DecisionAttribute.localValueCode(assignedDec)];
}
/**
* Returns string representation of classification results.
*
* @return String representation of classification results.
*/
public String toString()
{
StringBuffer sbuf = new StringBuffer();
if (m_Statistics!=null)
{
Enumeration enumarator = m_Statistics.propertyNames();
while (enumarator.hasMoreElements())
{
String resultName = (String)enumarator.nextElement();
sbuf.append(resultName+" = "+m_Statistics.getProperty(resultName)+Report.lineSeparator);
}
}
double accuracy = getAccuracy();
sbuf.append(" Accuracy: "+(int)(100*accuracy)+".");
if (10000*accuracy-100*(int)(100*accuracy) < 10) sbuf.append("0");
sbuf.append((int)(10000*accuracy-100*(int)(100*accuracy))+"%"+Report.lineSeparator);
for (int dec = 0; dec < m_DecisionAttribute.noOfValues(); dec++)
{
double decCode = m_DecisionAttribute.globalValueCode(dec);
double decAccuracy = getDecAccuracy(decCode);
sbuf.append(" Decision "+NominalAttribute.stringValue(decCode)+": "+(int)(100*decAccuracy)+".");
if (10000*decAccuracy-100*(int)(100*decAccuracy) < 10) sbuf.append("0");
sbuf.append((int)(10000*decAccuracy-100*(int)(100*decAccuracy))+"%"+Report.lineSeparator);
}
return sbuf.toString();
}
/**
* Returns statistics.
*
* @return Statistics.
*/
public Properties getStatistics()
{
Properties stats = (Properties)m_Statistics.clone();
stats.put("all_cnt",Integer.toString(m_nAll));
stats.put("cover_cnt",Integer.toString(m_nCovered));
stats.put("correct_cnt",Integer.toString(m_nCorrect));
stats.put("precision",Double.toString(((double)m_nCorrect)/((double)m_nCovered)));
stats.put("coverage",Double.toString(((double)m_nCovered)/((double)m_nAll)));
stats.put("accuracy",Double.toString(((double)m_nCorrect)/((double)m_nAll)));
return stats;
}
}
