decisiontable.java

代码是一个分类器的实现,其中使用了部分weka的源代码。可以将项目导入eclipse运行

/*
 *    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.
 */

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

package weka.classifiers.rules;

import weka.classifiers.Classifier;
import weka.classifiers.lazy.IBk;
import weka.classifiers.Evaluation;
import weka.core.AdditionalMeasureProducer;
import weka.core.Capabilities;
import weka.core.FastVector;
import weka.core.Instance;
import weka.core.Instances;
import weka.core.Option;
import weka.core.OptionHandler;
import weka.core.Tag;
import weka.core.TechnicalInformation;
import weka.core.TechnicalInformationHandler;
import weka.core.Utils;
import weka.core.WeightedInstancesHandler;
import weka.core.Capabilities.Capability;
import weka.core.TechnicalInformation.Field;
import weka.core.TechnicalInformation.Type;
import weka.core.SelectedTag;
import weka.filters.Filter;
import weka.filters.unsupervised.attribute.Remove;
import weka.attributeSelection.SubsetEvaluator;
import weka.attributeSelection.ASSearch;
import weka.attributeSelection.BestFirst;

import java.io.Serializable;
import java.util.Arrays;
import java.util.BitSet;
import java.util.Enumeration;
import java.util.Hashtable;
import java.util.Random;
import java.util.Vector;

/**
 <!-- globalinfo-start -->
 * Class for building and using a simple decision table majority classifier.<br/>
 * <br/>
 * For more information see: <br/>
 * <br/>
 * Ron Kohavi: The Power of Decision Tables. In: 8th European Conference on Machine Learning, 174-189, 1995.
 * <p/>
 <!-- globalinfo-end -->
 *
 <!-- technical-bibtex-start -->
 * BibTeX:
 * <pre>
 * &#64;inproceedings{Kohavi1995,
 *    author = {Ron Kohavi},
 *    booktitle = {8th European Conference on Machine Learning},
 *    pages = {174-189},
 *    publisher = {Springer},
 *    title = {The Power of Decision Tables},
 *    year = {1995}
 * }
 * </pre>
 * <p/>
 <!-- technical-bibtex-end -->
 *
 <!-- options-start -->
 * Valid options are: <p/>
 * 
 * <pre> -X &lt;number of folds&gt;
 *  Use cross validation to evaluate features.
 *  Use number of folds = 1 for leave one out CV.
 *  (Default = leave one out CV)</pre>
 * 
 * <pre> -I
 *  Use nearest neighbour instead of global table majority.</pre>
 * 
 * <pre> -R
 *  Display decision table rules.
 * </pre>
 * 
 <!-- options-end -->
 *
 * @author Mark Hall ([e-mail])
 * @version $Revision: 1.39 $ 
 */
public class DecisionTable 
extends Classifier 
implements OptionHandler, WeightedInstancesHandler, 
AdditionalMeasureProducer, TechnicalInformationHandler {

  /** for serialization */
  static final long serialVersionUID = 2888557078165701326L;
  //static final long serialVersionUID = 2888557078885701326L;

  /** The hashtable used to hold training instances */
  protected Hashtable m_entries;

  /** The class priors to use when there is no match in the table */
  protected double [] m_classPriorCounts;
  protected double [] m_classPriors;

  /** Holds the final feature set */
  protected int [] m_decisionFeatures;

  /** Discretization filter */
  protected Filter m_disTransform;

  /** Filter used to remove columns discarded by feature selection */
  protected Remove m_delTransform;

  /** IB1 used to classify non matching instances rather than majority class */
  protected IBk m_ibk;

  /** Holds the original training instances */
  protected Instances m_theInstances;

  /** Holds the final feature selected set of instances */
  protected Instances m_dtInstances;

  /** The number of attributes in the dataset */
  protected int m_numAttributes;

  /** The number of instances in the dataset */
  private int m_numInstances;

  /** Class is nominal */
  protected boolean m_classIsNominal;

  /** Use the IBk classifier rather than majority class */
  protected boolean m_useIBk;

  /** Display Rules */
  protected boolean m_displayRules;

  /** Number of folds for cross validating feature sets */
  private int m_CVFolds;

  /** Random numbers for use in cross validation */
  private Random m_rr;

  /** Holds the majority class */
  protected double m_majority;

  /** The search method to use */
  protected ASSearch m_search = new BestFirst();

  /** Our own internal evaluator */
  protected SubsetEvaluator m_evaluator;

  /** The evaluation object used to evaluate subsets */
  protected Evaluation m_evaluation;

  /** default is accuracy for discrete class and RMSE for numeric class */
  public static final int EVAL_DEFAULT = 1;
  public static final int EVAL_ACCURACY = 2;
  public static final int EVAL_RMSE = 3;
  public static final int EVAL_MAE = 4;
  public static final int EVAL_AUC = 5;

  public static final Tag [] TAGS_EVALUATION = {
    new Tag(EVAL_DEFAULT, "Default: accuracy (discrete class); RMSE (numeric class)"),
    new Tag(EVAL_ACCURACY, "Accuracy (discrete class only"),
    new Tag(EVAL_RMSE, "RMSE (of the class probabilities for discrete class)"),
    new Tag(EVAL_MAE, "MAE (of the class probabilities for discrete class)"),
    new Tag(EVAL_AUC, "AUC (area under the ROC curve - discrete class only)")
  };

  protected int m_evaluationMeasure = EVAL_DEFAULT;

  /**
   * Returns a string describing classifier
   * @return a description suitable for
   * displaying in the explorer/experimenter gui
   */
  public String globalInfo() {

    return  
    "Class for building and using a simple decision table majority "
    + "classifier.\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.INPROCEEDINGS);
    result.setValue(Field.AUTHOR, "Ron Kohavi");
    result.setValue(Field.TITLE, "The Power of Decision Tables");
    result.setValue(Field.BOOKTITLE, "8th European Conference on Machine Learning");
    result.setValue(Field.YEAR, "1995");
    result.setValue(Field.PAGES, "174-189");
    result.setValue(Field.PUBLISHER, "Springer");

    return result;
  }
  
  /**
   * Class providing keys to the hash table
   */
  public static class hashKey 
  implements Serializable {

    /** for serialization */
    static final long serialVersionUID = 5674163500154964602L;

    /** Array of attribute values for an instance */
    private double [] attributes;

    /** True for an index if the corresponding attribute value is missing. */
    private boolean [] missing;

    /** The key */
    private int key;

    /**
     * Constructor for a hashKey
     *
     * @param t an instance from which to generate a key
     * @param numAtts the number of attributes
     * @param ignoreClass if true treat the class as a normal attribute
     * @throws Exception if something goes wrong
     */
    public hashKey(Instance t, int numAtts, boolean ignoreClass) throws Exception {

      int i;
      int cindex = t.classIndex();

      key = -999;
      attributes = new double [numAtts];
      missing = new boolean [numAtts];
      for (i=0;i<numAtts;i++) {
	if (i == cindex && !ignoreClass) {
	  missing[i] = true;
	} else {
	  if ((missing[i] = t.isMissing(i)) == false) {
	    attributes[i] = t.value(i);
	  }
	}
      }
    }

    /**
     * Convert a hash entry to a string
     *
     * @param t the set of instances
     * @param maxColWidth width to make the fields
     * @return string representation of the hash entry
     */
    public String toString(Instances t, int maxColWidth) {

      int i;
      int cindex = t.classIndex();
      StringBuffer text = new StringBuffer();

      for (i=0;i<attributes.length;i++) {
	if (i != cindex) {
	  if (missing[i]) {
	    text.append("?");
	    for (int j=0;j<maxColWidth;j++) {
	      text.append(" ");
	    }
	  } else {
	    String ss = t.attribute(i).value((int)attributes[i]);
	    StringBuffer sb = new StringBuffer(ss);

	    for (int j=0;j < (maxColWidth-ss.length()+1); j++) {
	      sb.append(" ");
	    }
	    text.append(sb);
	  }
	}
      }
      return text.toString();
    }

    /**
     * Constructor for a hashKey
     *
     * @param t an array of feature values
     */
    public hashKey(double [] t) {

      int i;
      int l = t.length;

      key = -999;
      attributes = new double [l];
      missing = new boolean [l];
      for (i=0;i<l;i++) {
	if (t[i] == Double.MAX_VALUE) {
	  missing[i] = true;
	} else {
	  missing[i] = false;
	  attributes[i] = t[i];
	}
      }
    }

    /**
     * Calculates a hash code
     *
     * @return the hash code as an integer
     */
    public int hashCode() {

      int hv = 0;

      if (key != -999)
	return key;
      for (int i=0;i<attributes.length;i++) {
	if (missing[i]) {
	  hv += (i*13);
	} else {
	  hv += (i * 5 * (attributes[i]+1));
	}
      }
      if (key == -999) {
	key = hv;
      }
      return hv;
    }

    /**
     * Tests if two instances are equal
     *
     * @param b a key to compare with
     * @return true if both objects are equal
     */
    public boolean equals(Object b) {

      if ((b == null) || !(b.getClass().equals(this.getClass()))) {
	return false;
      }
      boolean ok = true;
      boolean l;
      if (b instanceof hashKey) {
	hashKey n = (hashKey)b;
	for (int i=0;i<attributes.length;i++) {
	  l = n.missing[i];
	  if (missing[i] || l) {
	    if ((missing[i] && !l) || (!missing[i] && l)) {
	      ok = false;
	      break;
	    }
	  } else {
	    if (attributes[i] != n.attributes[i]) {
	      ok = false;
	      break;
	    }
	  }
	}
      } else {
	return false;
      }
      return ok;
    }

    /**
     * Prints the hash code
     */
    public void print_hash_code() {

      System.out.println("Hash val: "+hashCode());
    }
  }

  /**
   * Inserts an instance into the hash table
   *
   * @param inst instance to be inserted
   * @param instA to create the hash key from
   * @throws Exception if the instance can't be inserted
   */
  private void insertIntoTable(Instance inst, double [] instA)
  throws Exception {

    double [] tempClassDist2;
    double [] newDist;
    hashKey thekey;

    if (instA != null) {
      thekey = new hashKey(instA);
    } else {
      thekey = new hashKey(inst, inst.numAttributes(), false);
    }

    // see if this one is already in the table
    tempClassDist2 = (double []) m_entries.get(thekey);
    if (tempClassDist2 == null) {
      if (m_classIsNominal) {
	newDist = new double [m_theInstances.classAttribute().numValues()];
	newDist[(int)inst.classValue()] = inst.weight();

	// add to the table
	m_entries.put(thekey, newDist);
      } else {
	newDist = new double [2];
	newDist[0] = inst.classValue() * inst.weight();
	newDist[1] = inst.weight();

	// add to the table
	m_entries.put(thekey, newDist);
      }
    } else { 

      // update the distribution for this instance
      if (m_classIsNominal) {
	tempClassDist2[(int)inst.classValue()]+=inst.weight();

	// update the table
	m_entries.put(thekey, tempClassDist2);
      } else  {
	tempClassDist2[0] += (inst.classValue() * inst.weight());
	tempClassDist2[1] += inst.weight();

	// update the table
	m_entries.put(thekey, tempClassDist2);
      }
    }
  }

  /**
   * Classifies an instance for internal leave one out cross validation
   * of feature sets
   *
   * @param instance instance to be "left out" and classified
   * @param instA feature values of the selected features for the instance
   * @return the classification of the instance
   * @throws Exception if something goes wrong
   */
  double evaluateInstanceLeaveOneOut(Instance instance, double [] instA)
  throws Exception {

    hashKey thekey;
    double [] tempDist;
    double [] normDist;

    thekey = new hashKey(instA);
    if (m_classIsNominal) {

      // if this one is not in the table
      if ((tempDist = (double [])m_entries.get(thekey)) == null) {
	throw new Error("This should never happen!");
      } else {
	normDist = new double [tempDist.length];
	System.arraycopy(tempDist,0,normDist,0,tempDist.length);
	normDist[(int)instance.classValue()] -= instance.weight();

	// update the table
	// first check to see if the class counts are all zero now
	boolean ok = false;
	for (int i=0;i<normDist.length;i++) {
	  if (!Utils.eq(normDist[i],0.0)) {
	    ok = true;
	    break;
	  }
	}

//	downdate the class prior counts
	m_classPriorCounts[(int)instance.classValue()] -= 
	  instance.weight();
	double [] classPriors = m_classPriorCounts.clone();
	Utils.normalize(classPriors);
	if (!ok) { // majority class
	  normDist = classPriors;
	}

	m_classPriorCounts[(int)instance.classValue()] += 
	  instance.weight();

	//if (ok) {
	Utils.normalize(normDist);
	if (m_evaluationMeasure == EVAL_AUC) {
	  m_evaluation.evaluateModelOnceAndRecordPrediction(normDist, instance);						
	} else {
	  m_evaluation.evaluateModelOnce(normDist, instance);
	}
	return Utils.maxIndex(normDist);