removeuselessattributes.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.preprocessing;

import edu.udo.cs.yale.operator.Operator;
import edu.udo.cs.yale.operator.IOObject;
import edu.udo.cs.yale.operator.OperatorException;
import edu.udo.cs.yale.example.Attribute;
import edu.udo.cs.yale.example.ExampleSet;
import edu.udo.cs.yale.operator.parameter.*;
import edu.udo.cs.yale.tools.LogService;

import java.util.List;
import java.util.Collection;
import java.util.Iterator;

/** Removes useless attribute from the example set. Useless attributes are
 *  <ul>
 *  <li>nominal attributes which has the same value for more than <code>p</code> percent of all examples.</li>
 *  <li>numerical attributes which standard deviation is less or equal to a given deviation threshold <code>t</code>.</li>
 *  </ul>
 *
 *  @version $Id: RemoveUselessAttributes.java,v 1.4 2004/08/28 17:04:49 ingomierswa Exp $
 */
public class RemoveUselessAttributes extends Operator {

    private static final Class[] INPUT_CLASSES = { ExampleSet.class };
    private static final Class[] OUTPUT_CLASSES = { ExampleSet.class };


    public IOObject[] apply() throws OperatorException {
	ExampleSet exampleSet = (ExampleSet)getInput(ExampleSet.class);
	ExampleSet clone = (ExampleSet)exampleSet.clone();

	double numericalMinDeviation  = getParameterAsDouble("numerical_min_deviation");
	double nominalMaxSingleValue = getParameterAsDouble("nominal_max_single_value");
	
	for (int i = 0; i < exampleSet.getNumberOfAttributes(); i++) {
	    Attribute attribute = exampleSet.getAttribute(i);

	    if (attribute.isNominal()) {
		Collection values = attribute.getValues();
		double[] valueCounts = new double[values.size()];
		Iterator v = values.iterator();
		int n = 0;
		while (v.hasNext()) {
		    String value = (String)v.next();
		    valueCounts[n] = attribute.getValueCount(value);
		    n++;
		}

		for (n = 0; n < valueCounts.length; n++) {
		    double percent = valueCounts[n] / (double)exampleSet.getSize();
		    if (percent >= nominalMaxSingleValue) {
			clone.removeAttribute(attribute);
			break;
		    }
		}
	    } else {
		if (Math.sqrt(attribute.getVariance()) <= numericalMinDeviation)
		    clone.removeAttribute(attribute);
	    }
	}

	if (clone.getNumberOfAttributes() <= 0) {
	    LogService.logMessage("Example set does not not have any attribute after removing the useless attributes!", LogService.WARNING);
	}

	return new IOObject[] { clone };
    }



    public Class[] getInputClasses() { return INPUT_CLASSES; }
    public Class[] getOutputClasses() { return OUTPUT_CLASSES; }    

    public List getParameterTypes() {
	List types = super.getParameterTypes();
	ParameterType type = new ParameterTypeDouble("numerical_min_deviation", 
						     "Removes all numerical attributes with standard deviation less or equal to this threshold.", 
						     0.0d, Double.POSITIVE_INFINITY, 0.0d);
	type.setExpert(false);
	types.add(type);
	type = new ParameterTypeDouble("nominal_max_single_value", 
				       "Removes all nominal attributes which provides more than the given amount of only one value.", 
				       0.0d, 1.0d, 1.0d);
	type.setExpert(false);
	types.add(type);
	return types;
    }

}