tertius.java

为了下东西 随便发了个 datamining 的源代码

   *
   * @return Tip text for this property suitable for
   * displaying in the explorer/experimenter GUI.
   */
  public String equivalentTipText() {
    
    return "Keep equivalent rules. "
      + "A rule r2 is equivalent to a rule r1 if the body of r2 is the "
      + "negation of the head of r1, and the head of r2 is the "
      + "negation of the body of r1.";
  }

  /**
   * Get the value of equivalent.
   *
   * @return Value of equivalent.
   */
  public boolean disabled_getEquivalent() {
    
    return !m_equivalent;
  }

  /**
   * Set the value of equivalent.
   *
   * @param v  Value to assign to equivalent.
   */
  public void disabled_setEquivalent(boolean v) {
    
    m_equivalent = !v;
  }

  /**
   * Returns the tip text for this property.
   *
   * @return Tip text for this property suitable for
   * displaying in the explorer/experimenter GUI.
   */
  public String sameClauseTipText() {

    return "Keep rules corresponding to the same clauses. "
      + "If set to false, only the rule with the best confirmation "
      + "value and rules with a lower number of counter-instances "
      + "will be kept.";
  }

  /**
   * Get the value of sameClause.
   *
   * @return Value of sameClause.
   */
  public boolean disabled_getSameClause() {

    return !m_sameClause;
  }

  /**
   * Set the value of sameClause.
   *
   * @param v  Value to assign to sameClause.
   */
  public void disabled_setSameClause(boolean v) {

    m_sameClause = !v;
  }

  /**
   * Returns the tip text for this property.
   *
   * @return Tip text for this property suitable for
   * displaying in the explorer/experimenter GUI.
   */
  public String subsumptionTipText() {

    return "Keep subsumed rules. "
      + "If set to false, subsumed rules will only be kept if they "
      + "have a better confirmation or a lower number of counter-instances.";
  }

  /**
   * Get the value of subsumption.
   *
   * @return Value of subsumption.
   */
  public boolean disabled_getSubsumption() {

    return !m_subsumption;
  }

  /**
   * Set the value of subsumption.
   *
   * @param v  Value to assign to subsumption.
   */
  public void disabled_setSubsumption(boolean v) {

    m_subsumption = !v;
  }

  /**
   * Returns the tip text for this property.
   *
   * @return Tip text for this property suitable for
   * displaying in the explorer/experimenter GUI.
   */
  public String missingValuesTipText() {
    
    return "Set the way to handle missing values. "
      + "Missing values can be set to match any value, or never match values "
      + "or to be significant and possibly appear in rules.";
  }

  /**
   * Get the value of missingValues.
   *
   * @return Value of missingValues.
   */
  public SelectedTag getMissingValues() {
    
    return new SelectedTag(m_missing, TAGS_MISSING);
  }

  /**
   * Set the value of missingValues.
   *
   * @param v  Value to assign to missingValues.
   */
  public void setMissingValues(SelectedTag v) {
    
    if (v.getTags() == TAGS_MISSING) {
      m_missing = v.getSelectedTag().getID();
    }
  }

  /**
   * Returns the tip text for this property.
   *
   * @return Tip text for this property suitable for
   * displaying in the explorer/experimenter GUI.
   */
  public String rocAnalysisTipText() {
    
    return "Return TP-rate and FP-rate for each rule found.";
  }

  /**
   * Get the value of rocAnalysis.
   *
   * @return Value of rocAnalysis.
   */
  public boolean getRocAnalysis() {

    return m_roc;
  }

  /**
   * Set the value of rocAnalysis.
   *
   * @param v  Value to assign to rocAnalysis.
   */
  public void setRocAnalysis(boolean v) {

    m_roc = v;
  }

  /**
   * Returns the tip text for this property.
   *
   * @return Tip text for this property suitable for
   * displaying in the explorer/experimenter GUI.
   */
  public String partFileTipText() {
    
    return "Set file containing the parts of the individual "
      + "for individual-based learning.";
  }

  /**
   * Get the value of partFile.
   *
   * @return Value of partFile.
   */
  public File disabled_getPartFile() {

    return new File(m_partsString);
  }

  /**
   * Set the value of partFile.
   *
   * @param v  Value to assign to partFile.
   */
  public void disabled_setPartFile(File v) throws Exception {

    m_partsString = v.getAbsolutePath();
    if (m_partsString.length() != 0) {
      Reader reader;
      try {
	reader = new BufferedReader(new FileReader(m_partsString));
      } catch (Exception e) {
	throw new Exception("Can't open file " + e.getMessage() + ".");
      }
      m_parts = new Instances(reader);	
    }
  }

  /**
   * Returns the tip text for this property.
   *
   * @return Tip text for this property suitable for
   * displaying in the explorer/experimenter GUI.
   */
  public String valuesOutputTipText() {
    
    return "Give visual feedback during the search. "
      + "The current best and worst values can be output either to stdout or to a separate window.";
  }

  /**
   * Get the value of valuesOutput.
   *
   * @return Value of valuesOutput.
   */
  public SelectedTag getValuesOutput() {
    
    return new SelectedTag(m_printValues, TAGS_VALUES);
  }

  /**
   * Set the value of valuesOutput.
   *
   * @param v  Value to assign to valuesOutput.
   */
  public void setValuesOutput(SelectedTag v) {
    
    if (v.getTags() == TAGS_VALUES) {
      m_printValues = v.getSelectedTag().getID();
    }
  }

  /**
   * Build the predicate corresponding to an attribute.
   *
   * @param instances The instances this predicates comes from.
   * @param attr The attribute this predicate corresponds to.
   * @param isClass Saying if the attribute is the class attribute.
   * @return The corresponding Predicate.
   * @exception Exception if the predicate could not be build 
   * (the attribute is numeric).
   */
  private Predicate buildPredicate(Instances instances,
				   Attribute attr, boolean isClass) 
    throws Exception {

    Predicate predicate; /* The result. */
    Literal lit;
    Literal negation;
    boolean missingValues; /* Missing values for this attribute ? */
    boolean individual = (m_parts != null); /* Individual-based learning ? */
    int type = (instances == m_parts)
      ? IndividualLiteral.PART_PROPERTY
      : IndividualLiteral.INDIVIDUAL_PROPERTY; /* Type of property. */

    if (attr.isNumeric()) {
      throw new Exception("Can't handle numeric attributes!");
    }
	
    missingValues = instances.attributeStats(attr.index()).missingCount > 0;

    /* Build predicate. */
    if (individual) {
      predicate = new Predicate(instances.relationName() + "." + attr.name(), 
				attr.index(), isClass);
    } else {
      predicate = new Predicate(attr.name(), attr.index(), isClass);
    }
	
    if (attr.numValues() == 2
	&& (!missingValues || m_missing == EXPLICIT)) {
      /* Case of two values.
       * If there are missing values, this case is treated like other cases.
       */
      if (individual) {
	lit = new IndividualLiteral(predicate, attr.value(0), 0, 
				    Literal.POS, m_missing, type);
	negation = new IndividualLiteral(predicate, attr.value(1), 1, 
					 Literal.POS, m_missing, type);
      } else {
	lit = new AttributeValueLiteral(predicate, attr.value(0), 0, 
					Literal.POS, m_missing);
	negation = new AttributeValueLiteral(predicate, attr.value(1), 1, 
					     Literal.POS, m_missing);
      }
      lit.setNegation(negation);
      negation.setNegation(lit);
      predicate.addLiteral(lit);      
    } else {
      /* Case of several values. */
      for (int i = 0; i < attr.numValues(); i++) {
	if (individual) {
	  lit = new IndividualLiteral(predicate, attr.value(i), i,
				      Literal.POS, m_missing, type);
	} else {
	  lit = new AttributeValueLiteral(predicate, attr.value(i), i, 
					  Literal.POS, m_missing);
	}
	if (m_negation != NONE) {
	  if (individual) {
	    negation = new IndividualLiteral(predicate, attr.value(i), i, 
					     Literal.NEG, m_missing, type);
	  } else {
	    negation = new AttributeValueLiteral(predicate, attr.value(i), i, 
						 Literal.NEG, m_missing);
	  }
	  lit.setNegation(negation);
	  negation.setNegation(lit);
	}
	predicate.addLiteral(lit);
      }

      /* One more value if missing is significant. */
      if (missingValues && m_missing == SIGNIFICANT) {
	if (individual) {
	  lit = new IndividualLiteral(predicate, "?", -1, 
				      Literal.POS, m_missing, type);
	} else {
	  lit = new AttributeValueLiteral(predicate, "?", -1, 
					  Literal.POS, m_missing);
	}
	if (m_negation != NONE) {
	  if (individual) {
	    negation = new IndividualLiteral(predicate, "?", -1, 
					     Literal.NEG, m_missing, type);
	  } else {
	    negation = new AttributeValueLiteral(predicate, "?", -1, 
						 Literal.NEG, m_missing);
	  }
	  lit.setNegation(negation);
	  negation.setNegation(lit);
	}
	predicate.addLiteral(lit);
      }
    }
    return predicate;
  }
   
  /*
   * Build the predicates to use in the rules.
   *
   * @exception If the predicates could not be built 
   * (numeric attribute).
   */
  private ArrayList buildPredicates() throws Exception {

    ArrayList predicates = new ArrayList(); /* The result. */
    Predicate predicate;
    Attribute attr;
    Enumeration attributes = m_instances.emerateAttributes();
    boolean individual = (m_parts != null); /* Individual-based learning ? */

    /* Attributes. */
    while (attributes.hasMoreElements()) {
      attr = (Attribute) attributes.nextElement();
      /* Identifiers do not appear in rules in individual-based learning. */
      if (!(individual && attr.name().equals("id"))) {
	predicate = buildPredicate(m_instances, attr, false);
	predicates.add(predicate);
      }
    }
    /* Class attribute. */
    attr = m_instances.classAttribute();
    /* Identifiers do not appear in rules. */
    if (!(individual && attr.name().equals("id"))) {
      predicate = buildPredicate(m_instances, attr, true);
      predicates.add(predicate);
    }

    /* Attributes of the parts in individual-based learning. */
    if (individual) {
      attributes = m_parts.emerateAttributes();
      while (attributes.hasMoreElements()) {
	attr = (Attribute) attributes.nextElement();
	/* Identifiers do not appear in rules. */
	if (!attr.name().equals("id")) {
	  predicate = buildPredicate(m_parts, attr, false);
	  predicates.add(predicate);
	}
      }
    }
	
    return predicates;
  }

  /**
   * Count the number of distinct confirmation values in the results.
   *
   * @return Number of confirmation values in the results.
   */
  private int numValuesInResult() {

    int result = 0;
    SimpleLinkedList.LinkedListIterator iter = m_results.iterator();
    Rule current;
    Rule next;
    if (!iter.hasNext()) {
      return result;
    } else {
      current = (Rule) iter.next();
      while (iter.hasNext()) {
	next = (Rule) iter.next();
	if (current.getConfirmation() > next.getConfirmation()) {
	  result++;
	}
	current = next;
      }
      return result + 1;
    }
  }

  /**
   * Test if it is worth refining a rule.
   *
   * @param rule The rule to consider.
   * @return True if the rule can be refined, false otherwise.
   */
  private boolean canRefine(Rule rule) {
    if (rule.isEmpty()) {
      return true;
    }
    if (m_best != 0) {
      if (numValuesInResult() < m_best) {
	return true;
      }
      Rule worstResult = (Rule) m_results.getLast();
      if (rule.getOptimistic() >= worstResult.getConfirmation()) {
	return true;
      }
      return false;
    } else {
      return true;
    }
  }

  /**
   * Test if it is worth calculating the optimistic estimate of a rule.
   *
   * @param rule The rule to consider.
   * @return True if the optimistic estimate can be calculated, false otherwise.
   */
  private boolean canCalculateOptimistic(Rule rule) {
    if (rule.hasTrueBody() || rule.hasFalseHead()) {
      return false;
    }
    if (!rule.overFrequencyThreshold(m_frequencyThreshold)) {
      return false;
    }
    return true;
  }

  /**
   * Test if a rule can be explored (if it is interesting for the results 
   * or for refining).
   *
   * @param rule The rule to consider.
   * @return True if the rule can be explored, false otherwise.
   */
  private boolean canExplore(Rule rule) {
    if (rule.getOptimistic() < m_confirmationThreshold) {
      return false;
    }
    if (m_best != 0) {
      if (numValuesInResult() < m_best) {
	return true;
      }