xvalidation.java

著名的开源仿真软件yale

/*
 *  YALE - Yet Another Learning Environment
 *  Copyright (C) 2002, 2003
 *      Simon Fischer, Ralf Klinkenberg, Ingo Mierswa, 
 *          Katharina Morik, Oliver Ritthoff
 *      Artificial Intelligence Unit
 *      Computer Science Department
 *      University of Dortmund
 *      44221 Dortmund,  Germany
 *  email: [e-mail]
 *  web:   http://yale.cs.uni-dortmund.de/
 *
 *  This program 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 2 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
 *  General Public License for more details.
 *
 *  You should have received a copy of the GNU General Public License
 *  along with this program; if not, write to the Free Software
 *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
 *  USA.
 */
package edu.udo.cs.yale.operator;

import edu.udo.cs.yale.operator.OperatorException;
import edu.udo.cs.yale.operator.parameter.*;
import edu.udo.cs.yale.example.ExampleSet;
import edu.udo.cs.yale.example.SplittedExampleSet;
import edu.udo.cs.yale.example.Example;
import edu.udo.cs.yale.example.ExampleReader;
import edu.udo.cs.yale.operator.performance.*;
import edu.udo.cs.yale.tools.*;
import java.util.List;


/** <code>XValidation</code> encapsulates a cross-validation experiment. The 
 *  example set {@yale.math S} is split up into <var> number_of_validations</var> subsets 
 *  {@yale.math S_i}. The inner operators are applied <var>number_of_validations</var>
 *  times using {@yale.math S_i} as the test set (input of the second inner operator) and 
 *  {@yale.math S\backslash S_i} training set (input of the first inner operator).
 *
 *  The first inner operator must accept an
 *  {@link edu.udo.cs.yale.example.ExampleSet} while the second must accept an
 *  {@link edu.udo.cs.yale.example.ExampleSet} and the output of the first (which is
 *  in most cases a {@link edu.udo.cs.yale.operator.learner.Model}) and must produce
 *  a {@link edu.udo.cs.yale.operator.performance.PerformanceVector}.
 *
 *  @yale.index cross-validation
 *  @yale.xmlclass XValidation
 *  @author  Ingo, Ralf
 *  @version $Id: XValidation.java,v 2.11 2003/07/09 15:19:58 fischer Exp $
 */
public class XValidation extends ValidationChain {
    
    private int number, iteration;
    
    public XValidation() {
	addValue(new Value("iteration", "The number of the current iteration.") {
		public double getValue() {
		    return iteration;
		}
	    });
    }

    public int getNumberOfValidationSteps() {
	return number;
    }


    public IOObject[] apply() throws OperatorException {
	ExampleSet inputSet = (ExampleSet)getInput(ExampleSet.class);
	if (getParameterAsBoolean("leave_one_out")) {
	    number = inputSet.getSize();
	} else {
	    number = getParameterAsInt("number_of_validations");
	}
	LogService.logMessage(getName() + ": Starting "+number+"-fold cross validation", LogService.TASK);
	PerformanceVector performanceVector = null;

	// Split training / test set
	SplittedExampleSet splittedES = new SplittedExampleSet(inputSet, number);

	// start crossvalidation
	for (iteration = 0; iteration < number; iteration++) {

	    splittedES.selectAllSubsetsBut(iteration);
	    learn(splittedES);

	    splittedES.selectSingleSubset(iteration);
	    IOContainer evalOutput = evaluate(splittedES);
	    PerformanceVector iterationPerformance = (PerformanceVector)evalOutput.getInput(PerformanceVector.class);  
	    if (performanceVector == null) {
		performanceVector = iterationPerformance;
	    } else {
		for (int i = 0; i  < performanceVector.size(); i++) {
		    performanceVector.get(i).buildAverage(iterationPerformance.get(i));
		}		     
	    }
	    setLastPerformance(iterationPerformance.getMainCriterion());

	    inApplyLoop();
	}
	// end crossvalidation

	setResult(performanceVector.getMainCriterion());
	return new IOObject[] { performanceVector };
    }


    public List getParameterTypes() {
	List types = super.getParameterTypes();
	types.add(new ParameterTypeInt("number_of_validations", "Number of subsets for the crossvalidation.", 2, Integer.MAX_VALUE, 10));
	types.add(new ParameterTypeBoolean("leave_one_out", "Set the number of validations to the number of examples. If set to true, number_of_validations is ignored.", false));
	return types;
    }
}