wrappersubseteval.java

来自「Java 编写的多种数据挖掘算法 包括聚类、分类、预处理等」· Java 代码 · 共 657 行 · 第 1/2 页

/*
 *    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., 675 Mass Ave, Cambridge, MA 02139, USA.
 */

/*
 *    WrapperSubsetEval.java
 *    Copyright (C) 1999 Mark Hall
 *
 */

package weka.attributeSelection;

import weka.classifiers.Classifier;
import weka.classifiers.Evaluation;
import weka.classifiers.rules.ZeroR;
import weka.core.Instances;
import weka.core.Option;
import weka.core.OptionHandler;
import weka.core.TechnicalInformation;
import weka.core.TechnicalInformation.Type;
import weka.core.TechnicalInformation.Field;
import weka.core.TechnicalInformationHandler;
import weka.core.UnsupportedAttributeTypeException;
import weka.core.Utils;
import weka.filters.Filter;
import weka.filters.unsupervised.attribute.Remove;

import java.util.BitSet;
import java.util.Enumeration;
import java.util.Random;
import java.util.Vector;

/** 
 <!-- globalinfo-start -->
 * WrapperSubsetEval:<br/>
 * <br/>
 * Evaluates attribute sets by using a learning scheme. Cross validation is used to estimate the accuracy of the learning scheme for a set of attributes.<br/>
 * <br/>
 * For more information see:<br/>
 * <br/>
 * Ron Kohavi, George H. John (1997). Wrappers for feature subset selection. Artificial Intelligence. 97(1-2):273-324.
 * <p/>
 <!-- globalinfo-end -->
 *
 <!-- technical-bibtex-start -->
 * BibTeX:
 * <pre>
 * &#64;article{Kohavi1997,
 *    author = {Ron Kohavi and George H. John},
 *    journal = {Artificial Intelligence},
 *    note = {Special issue on relevance},
 *    number = {1-2},
 *    pages = {273-324},
 *    title = {Wrappers for feature subset selection},
 *    volume = {97},
 *    year = {1997},
 *    ISSN = {0004-3702}
 * }
 * </pre>
 * <p/>
 <!-- technical-bibtex-end -->
 *
 <!-- options-start -->
 * Valid options are: <p/>
 * 
 * <pre> -B &lt;base learner&gt;
 *  class name of base learner to use for
 *  accuracy estimation. Place any
 *  classifier options LAST on the
 *  command line following a "--".
 *  eg. -B weka.classifiers.bayes.NaiveBayes ... -- -K</pre>
 * 
 * <pre> -F &lt;num&gt;
 *  number of cross validation folds to use
 *  for estimating accuracy.
 *  (default=5)</pre>
 * 
 * <pre> -R &lt;seed&gt;
 *  Seed for cross validation accuracy 
 *  estimation.
 *  (default = 1)</pre>
 * 
 * <pre> -T &lt;num&gt;
 *  threshold by which to execute another cross validation
 *  (standard deviation---expressed as a percentage of the mean).
 *  (default=0.01(1%))</pre>
 * 
 * <pre> 
 * Options specific to scheme weka.classifiers.rules.ZeroR:
 * </pre>
 * 
 * <pre> -D
 *  If set, classifier is run in debug mode and
 *  may output additional info to the console</pre>
 * 
 <!-- options-end -->
 *
 * @author Mark Hall ([e-mail])
 * @version $Revision: 1.26 $
 */
public class WrapperSubsetEval
  extends SubsetEvaluator
  implements OptionHandler, TechnicalInformationHandler {
  
  /** for serialization */
  static final long serialVersionUID = -4573057658746728675L;

  /** training instances */
  private Instances m_trainInstances;
  /** class index */
  private int m_classIndex;
  /** number of attributes in the training data */
  private int m_numAttribs;
  /** number of instances in the training data */
  private int m_numInstances;
  /** holds an evaluation object */
  private Evaluation m_Evaluation;
  /** holds the base classifier object */
  private Classifier m_BaseClassifier;
  /** number of folds to use for cross validation */
  private int m_folds;
  /** random number seed */
  private int m_seed;
  /** 
   * the threshold by which to do further cross validations when
   * estimating the accuracy of a subset
   */
  private double m_threshold;

  /**
   * Returns a string describing this attribute evaluator
   * @return a description of the evaluator suitable for
   * displaying in the explorer/experimenter gui
   */
  public String globalInfo() {
    return "WrapperSubsetEval:\n\n"
      +"Evaluates attribute sets by using a learning scheme. Cross "
      +"validation is used to estimate the accuracy of the learning "
      +"scheme for a set of attributes.\n\n"
      + "For more information see:\n\n"
      + getTechnicalInformation().toString();
  }

  /**
   * Returns an instance of a TechnicalInformation object, containing 
   * detailed information about the technical background of this class,
   * e.g., paper reference or book this class is based on.
   * 
   * @return the technical information about this class
   */
  public TechnicalInformation getTechnicalInformation() {
    TechnicalInformation 	result;
    
    result = new TechnicalInformation(Type.ARTICLE);
    result.setValue(Field.AUTHOR, "Ron Kohavi and George H. John");
    result.setValue(Field.YEAR, "1997");
    result.setValue(Field.TITLE, "Wrappers for feature subset selection");
    result.setValue(Field.JOURNAL, "Artificial Intelligence");
    result.setValue(Field.VOLUME, "97");
    result.setValue(Field.NUMBER, "1-2");
    result.setValue(Field.PAGES, "273-324");
    result.setValue(Field.NOTE, "Special issue on relevance");
    result.setValue(Field.ISSN, "0004-3702");
    
    return result;
  }

  /**
   * Constructor. Calls restOptions to set default options
   **/
  public WrapperSubsetEval () {
    resetOptions();
  }


  /**
   * Returns an enumeration describing the available options.
   * @return an enumeration of all the available options.
   **/
  public Enumeration listOptions () {
    Vector newVector = new Vector(4);
    newVector.addElement(new Option("\tclass name of base learner to use for" 
				    + "\n\taccuracy estimation. Place any" 
				    + "\n\tclassifier options LAST on the" 
				    + "\n\tcommand line following a \"--\"." 
				    + "\n\teg. -B weka.classifiers.bayes.NaiveBayes ... " 
				    + "-- -K", "B", 1, "-B <base learner>"));
    newVector.addElement(new Option("\tnumber of cross validation folds to " 
				    + "use\n\tfor estimating accuracy." 
				    + "\n\t(default=5)", "F", 1, "-F <num>"));
    newVector.addElement(new Option("\tSeed for cross validation accuracy "
				    +"\n\testimation."
				    +"\n\t(default = 1)", "R", 1,"-R <seed>"));
    newVector.addElement(new Option("\tthreshold by which to execute " 
				    + "another cross validation" 
				    + "\n\t(standard deviation---" 
				    + "expressed as a percentage of the " 
				    + "mean).\n\t(default=0.01(1%))"
				    , "T", 1, "-T <num>"));

    if ((m_BaseClassifier != null) && 
	(m_BaseClassifier instanceof OptionHandler)) {
      newVector.addElement(new Option("", "", 0, "\nOptions specific to scheme " 
				      + m_BaseClassifier.getClass().getName() 
				      + ":"));
      Enumeration enu = ((OptionHandler)m_BaseClassifier).listOptions();

      while (enu.hasMoreElements()) {
        newVector.addElement(enu.nextElement());
      }
    }

    return  newVector.elements();
  }


  /**
   * Parses a given list of options. <p/>
   *
   <!-- options-start -->
   * Valid options are: <p/>
   * 
   * <pre> -B &lt;base learner&gt;
   *  class name of base learner to use for
   *  accuracy estimation. Place any
   *  classifier options LAST on the
   *  command line following a "--".
   *  eg. -B weka.classifiers.bayes.NaiveBayes ... -- -K</pre>
   * 
   * <pre> -F &lt;num&gt;
   *  number of cross validation folds to use
   *  for estimating accuracy.
   *  (default=5)</pre>
   * 
   * <pre> -R &lt;seed&gt;
   *  Seed for cross validation accuracy 
   *  estimation.
   *  (default = 1)</pre>
   * 
   * <pre> -T &lt;num&gt;
   *  threshold by which to execute another cross validation
   *  (standard deviation---expressed as a percentage of the mean).
   *  (default=0.01(1%))</pre>
   * 
   * <pre> 
   * Options specific to scheme weka.classifiers.rules.ZeroR:
   * </pre>
   * 
   * <pre> -D
   *  If set, classifier is run in debug mode and
   *  may output additional info to the console</pre>
   * 
   <!-- options-end -->
   *
   * @param options the list of options as an array of strings
   * @throws Exception if an option is not supported
   */
  public void setOptions (String[] options)
    throws Exception {
    String optionString;
    resetOptions();
    optionString = Utils.getOption('B', options);

    if (optionString.length() == 0) {
      throw  new Exception("A learning scheme must be specified with" 
			   + "-B option");
    }

    setClassifier(Classifier.forName(optionString, 
				     Utils.partitionOptions(options)));
    optionString = Utils.getOption('F', options);

    if (optionString.length() != 0) {
      setFolds(Integer.parseInt(optionString));
    }

    optionString = Utils.getOption('R', options);
    if (optionString.length() != 0) {
      setSeed(Integer.parseInt(optionString));
    }

    //       optionString = Utils.getOption('S',options);
    //       if (optionString.length() != 0)
    //         {
    //  	 seed = Integer.parseInt(optionString);
    //         }
    optionString = Utils.getOption('T', options);

    if (optionString.length() != 0) {
      Double temp;
      temp = Double.valueOf(optionString);
      setThreshold(temp.doubleValue());
    }
  }
  
  /**
   * Returns the tip text for this property
   * @return tip text for this property suitable for
   * displaying in the explorer/experimenter gui
   */
  public String thresholdTipText() {
    return "Repeat xval if stdev of mean exceeds this value.";
  }

  /**
   * Set the value of the threshold for repeating cross validation
   *
   * @param t the value of the threshold
   */
  public void setThreshold (double t) {
    m_threshold = t;
  }


  /**
   * Get the value of the threshold
   *