xvalidation.java

一个很好的LIBSVM的JAVA源码。对于要研究和改进SVM算法的学者。可以参考。来自数据挖掘工具YALE工具包。

/*
 *  YALE - Yet Another Learning Environment
 *  Copyright (C) 2001-2004
 *      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.validation;

import edu.udo.cs.yale.operator.IOObject;
import edu.udo.cs.yale.operator.IOContainer;
import edu.udo.cs.yale.operator.Value;
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.math.*;
import edu.udo.cs.yale.tools.LogService;

import java.util.List;
import java.util.ArrayList;


/** <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 1.4 2004/10/07 20:31:00 ingomierswa 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);

	// Split training / test set
        boolean shuffle = getParameterAsBoolean("shuffle");
	SplittedExampleSet splittedES = new SplittedExampleSet(inputSet, number, shuffle);

	// start crossvalidation
	List averageVectors = new ArrayList();
	for (iteration = 0; iteration < number; iteration++) {

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

	    splittedES.selectSingleSubset(iteration);
	    IOContainer evalOutput = evaluate(splittedES);
	    handleAverages(evalOutput, averageVectors);

	    inApplyLoop();
	}
	// end crossvalidation

	PerformanceVector averagePerformance = getPerformanceVector(averageVectors);
 	if (averagePerformance != null)
 	    setResult(averagePerformance.getMainCriterion());
	
	AverageVector[] result = new AverageVector[averageVectors.size()];
	averageVectors.toArray(result);
	return result;
    }

    public List getParameterTypes() {
	List types = super.getParameterTypes();
	ParameterType type = new ParameterTypeInt("number_of_validations", "Number of subsets for the crossvalidation.", 2, Integer.MAX_VALUE, 10);
	type.setExpert(false);
	types.add(type);
	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));
        types.add(new ParameterTypeBoolean("shuffle", "If true, shuffle the dataset records.", true));
	return types;
    }
}