tertius.java

Java 编写的多种数据挖掘算法 包括聚类、分类、预处理等

/*
 *    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.
 */

/*
 *    Tertius.java
 *    Copyright (C) 2003 Peter A. Flach, Nicolas Lachiche
 *
 *    Thanks to Amelie Deltour for porting the original C code to Java
 *    and integrating it into Weka.
 */

package weka.associations;

import weka.associations.tertius.AttributeValueLiteral;
import weka.associations.tertius.IndividualInstances;
import weka.associations.tertius.IndividualLiteral;
import weka.associations.tertius.Literal;
import weka.associations.tertius.Predicate;
import weka.associations.tertius.Rule;
import weka.associations.tertius.SimpleLinkedList;
import weka.core.Attribute;
import weka.core.Capabilities;
import weka.core.Instances;
import weka.core.Option;
import weka.core.OptionHandler;
import weka.core.SelectedTag;
import weka.core.Tag;
import weka.core.TechnicalInformation;
import weka.core.Capabilities.Capability;
import weka.core.TechnicalInformation.Type;
import weka.core.TechnicalInformation.Field;
import weka.core.TechnicalInformationHandler;
import weka.core.Utils;

import java.awt.BorderLayout;
import java.awt.Button;
import java.awt.Font;
import java.awt.Frame;
import java.awt.Label;
import java.awt.TextField;
import java.awt.event.ActionEvent;
import java.awt.event.ActionListener;
import java.io.BufferedReader;
import java.io.File;
import java.io.FileReader;
import java.io.Reader;
import java.text.SimpleDateFormat;
import java.util.ArrayList;
import java.util.Date;
import java.util.Enumeration;
import java.util.Vector;

/**
 <!-- globalinfo-start -->
 * Finds rules according to confirmation measure (Tertius-type algorithm).<br/>
 * <br/>
 * For more information see:<br/>
 * <br/>
 * P. A. Flach, N. Lachiche (1999). Confirmation-Guided Discovery of first-order rules with Tertius. Machine Learning. 42:61-95.
 * <p/>
 <!-- globalinfo-end -->
 * 
 <!-- technical-bibtex-start -->
 * BibTeX:
 * <pre>
 * &#64;article{Flach1999,
 *    author = {P. A. Flach and N. Lachiche},
 *    journal = {Machine Learning},
 *    pages = {61-95},
 *    title = {Confirmation-Guided Discovery of first-order rules with Tertius},
 *    volume = {42},
 *    year = {1999}
 * }
 * </pre>
 * <p/>
 <!-- technical-bibtex-end -->
 *
 <!-- options-start -->
 * Valid options are: <p/>
 * 
 * <pre> -K &lt;number of values in result&gt;
 *  Set maximum number of confirmation  values in the result. (default: 10)</pre>
 * 
 * <pre> -F &lt;frequency threshold&gt;
 *  Set frequency threshold for pruning. (default: 0)</pre>
 * 
 * <pre> -C &lt;confirmation threshold&gt;
 *  Set confirmation threshold. (default: 0)</pre>
 * 
 * <pre> -N &lt;noise threshold&gt;
 *  Set noise threshold : maximum frequency of counter-examples.
 *  0 gives only satisfied rules. (default: 1)</pre>
 * 
 * <pre> -R
 *  Allow attributes to be repeated in a same rule.</pre>
 * 
 * <pre> -L &lt;number of literals&gt;
 *  Set maximum number of literals in a rule. (default: 4)</pre>
 * 
 * <pre> -G &lt;0=no negation | 1=body | 2=head | 3=body and head&gt;
 *  Set the negations in the rule. (default: 0)</pre>
 * 
 * <pre> -S
 *  Consider only classification rules.</pre>
 * 
 * <pre> -c &lt;class index&gt;
 *  Set index of class attribute. (default: last).</pre>
 * 
 * <pre> -H
 *  Consider only horn clauses.</pre>
 * 
 * <pre> -E
 *  Keep equivalent rules.</pre>
 * 
 * <pre> -M
 *  Keep same clauses.</pre>
 * 
 * <pre> -T
 *  Keep subsumed rules.</pre>
 * 
 * <pre> -I &lt;0=always match | 1=never match | 2=significant&gt;
 *  Set the way to handle missing values. (default: 0)</pre>
 * 
 * <pre> -O
 *  Use ROC analysis. </pre>
 * 
 * <pre> -p &lt;name of file&gt;
 *  Set the file containing the parts of the individual for individual-based learning.</pre>
 * 
 * <pre> -P &lt;0=no output | 1=on stdout | 2=in separate window&gt;
 *  Set output of current values. (default: 0)</pre>
 * 
 <!-- options-end -->
 *
 * @author <a href="mailto:[e-mail]">Amelie Deltour</a>
 * @version $Revision: 1.6 $
 */

public class Tertius 
  extends Associator 
  implements OptionHandler, Runnable, TechnicalInformationHandler {

  /** for serialization */
  static final long serialVersionUID = 5556726848380738179L;
  
  /** The results. */
  private SimpleLinkedList m_results;

  /** Number of hypotheses considered. */
  private int m_hypotheses;

  /** Number of hypotheses explored. */
  private int m_explored;

  /** Time needed for the search. */
  private Date m_time;

  /** Field to output the current values. */ 
  private TextField m_valuesText;

  /** Instances used for the search. */
  private Instances m_instances;

  /** Predicates used in the rules. */
  private ArrayList m_predicates;

  /** Status of the search. */
  private int m_status;
  /** Status of the search: normal */
  private static final int NORMAL = 0;
  /** Status of the search: memory problem */
  private static final int MEMORY = 1;
  /** Status of the search: user interruption */
  private static final int STOP = 2;
  
  /* Pruning options. */

  /** Number of best confirmation values to search. */
  private int m_best;

  /** Frequency threshold for the body and the negation of the head. */
  private double m_frequencyThreshold;

  /** Confirmation threshold for the rules. */
  private double m_confirmationThreshold;

  /** Maximal number of counter-instances. */
  private double m_noiseThreshold;

  /* Search space & language bias options. */

  /** Repeat attributes ? */
  private boolean m_repeat;

  /** Number of literals in a rule. */
  private int m_numLiterals;

  /** Type of negation: none */
  private static final int NONE = 0;
  /** Type of negation: body */
  private static final int BODY = 1;
  /** Type of negation: head */
  private static final int HEAD = 2;
  /** Type of negation: all */
  private static final int ALL = 3;
  /** Types of negation. */
  private static final Tag [] TAGS_NEGATION = {
    new Tag(NONE, "None"),
    new Tag(BODY, "Body"),
    new Tag(HEAD, "Head"),
    new Tag(ALL, "Both")
      };

  /** Type of negation used in the rules. */
  private int m_negation;

  /** Classification bias. */
  private boolean m_classification;

  /** Index of class attribute. */
  private int m_classIndex;

  /** Horn clauses bias. */
  private boolean m_horn;

  /* Subsumption tests options. */

  /** Perform test on equivalent rules ? */
  private boolean m_equivalent;

  /** Perform test on same clauses ? */
  private boolean m_sameClause;
  
  /** Perform subsumption test ? */
  private boolean m_subsumption;

  /** Way of handling missing values: min counterinstances */
  public static final int EXPLICIT = 0;
  /** Way of handling missing values: max counterinstances */
  public static final int IMPLICIT = 1;
  /** Way of handling missing values: missing as a particular value */
  public static final int SIGNIFICANT = 2;
  /** Ways of handling missing values.  */
  private static final Tag [] TAGS_MISSING = {
    new Tag(EXPLICIT, "Matches all"),
    new Tag(IMPLICIT, "Matches none"),
    new Tag(SIGNIFICANT, "Significant")
      };

  /** Way of handling missing values in the search. */
  private int m_missing;

  /** Perform ROC analysis ? */
  private boolean m_roc;

  /** Name of the file containing the parts for individual-based learning. */
  private String m_partsString;
  
  /** Part instances for individual-based learning. */
  private Instances m_parts;

  /** Type of values output: No */ 
  private static final int NO = 0;
  /** Type of values output: stdout */ 
  private static final int OUT = 1;
  /** Type of values output: Window */ 
  private static final int WINDOW = 2;
  /** Types of values output. */ 
  private static final Tag [] TAGS_VALUES = {
    new Tag(NO, "No"),
    new Tag(OUT, "stdout"),
    new Tag(WINDOW, "Window")
      };

  /** Type of values output. */
  private int m_printValues;

  /**
   * Constructor that sets the options to the default values.
   */
  public Tertius() {

    resetOptions();
  }

  /**
   * Returns a string describing this associator.
   *
   * @return A description of the evaluator suitable for
   * displaying in the explorer/experimenter gui.
   */
  public String globalInfo() {
    return 
        "Finds rules according to confirmation measure (Tertius-type "
      + "algorithm).\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, "P. A. Flach and N. Lachiche");
    result.setValue(Field.YEAR, "1999");
    result.setValue(Field.TITLE, "Confirmation-Guided Discovery of first-order rules with Tertius");
    result.setValue(Field.JOURNAL, "Machine Learning");
    result.setValue(Field.VOLUME, "42");
    result.setValue(Field.PAGES, "61-95");
    
    return result;
  }

  /**
   * Resets the options to the default values.
   */
  public void resetOptions() {

    /* Pruning options. */
    m_best = 10;
    m_frequencyThreshold = 0;
    m_confirmationThreshold = 0;
    m_noiseThreshold = 1;

    /* Search space & language bias options. */
    m_repeat = false;
    m_numLiterals = 4;
    m_negation = NONE;
    m_classification = false;
    m_classIndex = 0;
    m_horn = false;

    /* Subsumption tests options. */
    m_equivalent = true;
    m_sameClause = true;
    m_subsumption = true;

    /* Missing values. */
    m_missing = EXPLICIT;

    /* ROC analysis. */
    m_roc = false;

    /* Individual-based learning. */
    m_partsString = "";
    m_parts = null;

    /* Values output. */
    m_printValues = NO;
  }

  /**
   * Returns an enumeration describing the available options.
   *
   * @return An enumeration of all the available options.
   */
  public Enumeration listOptions() {
    
    Vector newVector = new Vector(17);

    /* Pruning options. */
    newVector.addElement(new Option("\tSet maximum number of confirmation  "
				    + "values in the result. (default: 10)",
				    "K", 1, "-K <number of values in result>"));
    newVector.addElement(new Option("\tSet frequency threshold for pruning. "
				    + "(default: 0)",
				    "F", 1, "-F <frequency threshold>"));
    newVector.addElement(new Option("\tSet confirmation threshold. "
				    + "(default: 0)",
				    "C", 1, "-C <confirmation threshold>"));
    newVector.addElement(new Option("\tSet noise threshold : maximum frequency "
				    + "of counter-examples.\n\t0 gives only "
				    + "satisfied rules. (default: 1)",
				    "N", 1, "-N <noise threshold>"));

    /* Search space & language bias options. */
    newVector.addElement(new Option("\tAllow attributes to be repeated in a "
				    + "same rule.",
				    "R", 0, "-R"));
    newVector.addElement(new Option("\tSet maximum number of literals in a "
				    + "rule. (default: 4)",
				    "L", 1, "-L <number of literals>"));
    newVector.addElement(new Option("\tSet the negations in the rule. "
				    + "(default: 0)",
				    "G", 1, "-G <0=no negation | "
				    + "1=body | "
				    + "2=head | "
				    + "3=body and head>"));
    newVector.addElement(new Option("\tConsider only classification rules.",
				    "S", 0, "-S"));
    newVector.addElement(new Option("\tSet index of class attribute. "
				    + "(default: last).",
				    "c", 1, "-c <class index>"));
    newVector.addElement(new Option("\tConsider only horn clauses.",
				    "H", 0, "-H"));

    /* Subsumption tests options. */
    newVector.addElement(new Option("\tKeep equivalent rules.",
				    "E", 0, "-E"));
    newVector.addElement(new Option("\tKeep same clauses.",
				    "M", 0, "-M"));
    newVector.addElement(new Option("\tKeep subsumed rules.",
				    "T", 0, "-T"));

    /* Missing values options. */
    newVector.addElement(new Option("\tSet the way to handle missing values. " 
				    + "(default: 0)",
				    "I", 1, "-I <0=always match | "
				    + "1=never match | "
				    + "2=significant>"));

    /* ROC analysis. */
    newVector.addElement(new Option("\tUse ROC analysis. ",
				    "O", 0, "-O"));

    /* Individual-based learning. */
    newVector.addElement(new Option("\tSet the file containing the parts of "
				    + "the individual for individual-based "
				    + "learning.",
				    "p", 1, "-p <name of file>"));

    /* Values output. */
    newVector.addElement(new Option("\tSet output of current values. "
				    + "(default: 0)",
				    "P", 1, "-P <0=no output | "
				    + "1=on stdout | "
				    + "2=in separate window>"));
    
    return newVector.elements();
  }
  
  /**
   * Parses a given list of options. <p/>
   *
   <!-- options-start -->
   * Valid options are: <p/>
   * 
   * <pre> -K &lt;number of values in result&gt;
   *  Set maximum number of confirmation  values in the result. (default: 10)</pre>
   * 
   * <pre> -F &lt;frequency threshold&gt;
   *  Set frequency threshold for pruning. (default: 0)</pre>
   * 
   * <pre> -C &lt;confirmation threshold&gt;
   *  Set confirmation threshold. (default: 0)</pre>
   * 
   * <pre> -N &lt;noise threshold&gt;
   *  Set noise threshold : maximum frequency of counter-examples.
   *  0 gives only satisfied rules. (default: 1)</pre>
   * 
   * <pre> -R
   *  Allow attributes to be repeated in a same rule.</pre>
   * 
   * <pre> -L &lt;number of literals&gt;
   *  Set maximum number of literals in a rule. (default: 4)</pre>
   * 
   * <pre> -G &lt;0=no negation | 1=body | 2=head | 3=body and head&gt;
   *  Set the negations in the rule. (default: 0)</pre>
   * 
   * <pre> -S
   *  Consider only classification rules.</pre>
   * 
   * <pre> -c &lt;class index&gt;
   *  Set index of class attribute. (default: last).</pre>
   * 
   * <pre> -H
   *  Consider only horn clauses.</pre>
   * 
   * <pre> -E
   *  Keep equivalent rules.</pre>
   * 
   * <pre> -M
   *  Keep same clauses.</pre>
   * 
   * <pre> -T
   *  Keep subsumed rules.</pre>
   * 
   * <pre> -I &lt;0=always match | 1=never match | 2=significant&gt;
   *  Set the way to handle missing values. (default: 0)</pre>
   * 
   * <pre> -O
   *  Use ROC analysis. </pre>
   * 
   * <pre> -p &lt;name of file&gt;
   *  Set the file containing the parts of the individual for individual-based learning.</pre>
   * 
   * <pre> -P &lt;0=no output | 1=on stdout | 2=in separate window&gt;
   *  Set output of current values. (default: 0)</pre>
   *