predictiveapriori.java

Java 编写的多种数据挖掘算法 包括聚类、分类、预处理等

     * Gets the class attribute of all instances
     * @return Instances containing only the class attribute
     */    
  public Instances getInstancesOnlyClass() {
      
      Instances onlyClass = null;
      try{
        onlyClass = LabeledItemSet.divide(m_instances,true);
      } 
      catch(Exception e){
        e.printStackTrace();
        System.out.println("\n"+e.getMessage());
      }
      return onlyClass;
      
  }  

  /**
   * Returns an enumeration describing the available options.
   *
   * @return an enumeration of all the available options.
   */
  public Enumeration listOptions() {

    String string1 = "\tThe required number of rules. (default = " + (m_numRules-5) + ")",
      string2 = "\tIf set class association rules are mined. (default = no)",
      string3 = "\tThe class index. (default = last)";
    FastVector newVector = new FastVector(3);

    newVector.addElement(new Option(string1, "N", 1, 
				    "-N <required number of rules output>"));
    newVector.addElement(new Option(string2, "A", 0,
				    "-A"));
    newVector.addElement(new Option(string3, "c", 1,
				    "-c <the class index>"));
    return newVector.elements();
  }

 
/**
   * Parses a given list of options. <p/>
   * 
   <!-- options-start -->
   * Valid options are: <p/>
   * 
   * <pre> -N &lt;required number of rules output&gt;
   *  The required number of rules. (default = 100)</pre>
   * 
   * <pre> -A
   *  If set class association rules are mined. (default = no)</pre>
   * 
   * <pre> -c &lt;the class index&gt;
   *  The class index. (default = last)</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 {
    
    resetOptions();
    String numRulesString = Utils.getOption('N', options),
           classIndexString = Utils.getOption('c',options);
    
    if (numRulesString.length() != 0) 
	m_numRules = Integer.parseInt(numRulesString)+5;
    else
        m_numRules = Integer.MAX_VALUE;
    if (classIndexString.length() != 0) 
	m_classIndex = Integer.parseInt(numRulesString);
    m_car = Utils.getFlag('A', options);
  }

  /**
   * Gets the current settings of the PredictiveApriori object.
   *
   * @return an array of strings suitable for passing to setOptions
   */
  public String [] getOptions() {

    String [] options = new String [10];
    int current = 0;
    options[current++] = "-N"; options[current++] = "" + (m_numRules-5);
    options[current++] = "-A"; options[current++] = "" + m_car;
    options[current++] = "-c"; options[current++] = "" + m_classIndex;
    while (current < options.length) {
      options[current++] = "";
    }
    return options;
  }


  /**
   * Outputs the association rules.
   * 
   * @return a string representation of the model
   */
  public String toString() {

    StringBuffer text = new StringBuffer();

    if (m_allTheRules[0].size() == 0)
      return "\nNo large itemsets and rules found!\n";
    text.append("\nPredictiveApriori\n===================\n\n");
    text.append("\nBest rules found:\n\n");
    
    for (int i = 0; i < m_allTheRules[0].size(); i++) {
	    text.append(Utils.doubleToString((double)i+1, 
					     (int)(Math.log(m_numRules)/Math.log(10)+1),0)+
			". " + ((ItemSet)m_allTheRules[0].elementAt(i)).
			toString(m_instances) 
			+ " ==> " + ((ItemSet)m_allTheRules[1].elementAt(i)).
			toString(m_instances) +"    acc:("+  
			Utils.doubleToString(((Double)m_allTheRules[2].
					      elementAt(i)).doubleValue(),5)+")");
      
      text.append('\n');
    }
    
    
    return text.toString();
  }

 
  /**
   * Returns the tip text for this property
   * @return tip text for this property suitable for
   * displaying in the explorer/experimenter gui
   */
  public String numRulesTipText() {
    return "Number of rules to find.";
  }

  /**
   * Get the value of the number of required rules.
   *
   * @return Value of the number of required rules.
   */
  public int getNumRules() {
      
    return m_numRules-5;
  }
  
  /**
   * Set the value of required rules.
   *
   * @param v  Value to assign to number of required rules.
   */
  public void setNumRules(int v) {
	 
      m_numRules = v+5;
  }
  
    /**
   * Sets the class index
   * @param index the index of the class attribute
   */  
  public void setClassIndex(int index){
      
      m_classIndex = index;
  }
  
  /**
   * Gets the index of the class attribute
   * @return the index of the class attribute
   */  
  public int getClassIndex(){
      
      return m_classIndex;
  }

  /**
   * Returns the tip text for this property
   * @return tip text for this property suitable for
   * displaying in the explorer/experimenter gui
   */
  public String classIndexTipText() {
    return "Index of the class attribute.\n If set to -1, the last attribute will be taken as the class attribute.";
  }

    /**
   * Sets class association rule mining
   * @param flag if class association rules are mined, false otherwise
   */  
  public void setCar(boolean flag){
      
      m_car = flag;
  }
  
  /**
   * Gets whether class association ruels are mined
   * @return true if class association rules are mined, false otherwise
   */  
  public boolean getCar(){
      
      return m_car;
  }

  /**
   * Returns the tip text for this property
   * @return tip text for this property suitable for
   * displaying in the explorer/experimenter gui
   */
  public String carTipText() {
    return "If enabled class association rules are mined instead of (general) association rules.";
  }
  
    /**
   * Returns the metric string for the chosen metric type.
   * Predictive apriori uses the estimated predictive accuracy.
   * Therefore the metric string is "acc".
   * @return string "acc"
   */
  public String metricString() {
      
      return "acc";
  }

 
  /** 
   * Method that finds all large itemsets for the given set of instances.
   *
   * @param index the instances to be used
   * @throws Exception if an attribute is numeric
   */
  private void findLargeItemSets(int index) throws Exception {
    
    FastVector kMinusOneSets, kSets = new FastVector();
    Hashtable hashtable;
    int i = 0;
    // Find large itemsets
    //of length 1
    if(index == 1){
        kSets = ItemSet.singletons(m_instances);
        ItemSet.upDateCounters(kSets, m_instances);
        kSets = ItemSet.deleteItemSets(kSets, m_premiseCount,Integer.MAX_VALUE);
        if (kSets.size() == 0)
            return;
        m_Ls.addElement(kSets);
    }
    //of length > 1
    if(index >1){
        if(m_Ls.size() > 0)
            kSets = (FastVector)m_Ls.lastElement();
        m_Ls.removeAllElements();
        i = index-2;
        kMinusOneSets = kSets;
        kSets = ItemSet.mergeAllItemSets(kMinusOneSets, i, m_instances.numInstances());
        hashtable = ItemSet.getHashtable(kMinusOneSets, kMinusOneSets.size());
        m_hashtables.addElement(hashtable);
        kSets = ItemSet.pruneItemSets(kSets, hashtable);
        ItemSet.upDateCounters(kSets, m_instances);
        kSets = ItemSet.deleteItemSets(kSets, m_premiseCount,Integer.MAX_VALUE);
        if(kSets.size() == 0)
            return;
        m_Ls.addElement(kSets);
    }
  } 


  

  /** 
   * Method that finds all association rules.
   *
   * @throws Exception if an attribute is numeric
   */
  private void findRulesQuickly() throws Exception {

    RuleGeneration currentItemSet;
    
    // Build rules
    for (int j = 0; j < m_Ls.size(); j++) {
      FastVector currentItemSets = (FastVector)m_Ls.elementAt(j);
      Enumeration enumItemSets = currentItemSets.elements();
      while (enumItemSets.hasMoreElements()) { 
        currentItemSet = new RuleGeneration((ItemSet)enumItemSets.nextElement());
        m_best = currentItemSet.generateRules(m_numRules, m_midPoints,m_priors,m_expectation,
                                        m_instances,m_best,m_count);
          
        m_count = currentItemSet.m_count;
        if(!m_bestChanged && currentItemSet.m_change)
           m_bestChanged = true;
        //update minimum expected predictive accuracy to get into the n best
        if(m_best.size() >0)
            m_expectation = ((RuleItem)m_best.first()).accuracy();
        else m_expectation =0;
      }
    }
  }
  
  
  /**
   * Method that finds all large itemsets for class association rule mining for the given set of instances.
   * @param index the size of the large item sets
   * @throws Exception if an attribute is numeric
   */
  private void findLargeCarItemSets(int index) throws Exception {
    
    FastVector kMinusOneSets, kSets = new FastVector();
    Hashtable hashtable;
    int i = 0;
    // Find large itemsets
    if(index == 1){
        kSets = CaRuleGeneration.singletons(m_instances);
        ItemSet.upDateCounters(kSets, m_instances);
        kSets = ItemSet.deleteItemSets(kSets, m_premiseCount,Integer.MAX_VALUE);
        if (kSets.size() == 0)
            return;
        m_Ls.addElement(kSets);
    }
    
    if(index >1){
        if(m_Ls.size() > 0)
            kSets = (FastVector)m_Ls.lastElement();
        m_Ls.removeAllElements();
        i = index-2;
        kMinusOneSets = kSets;
        kSets = ItemSet.mergeAllItemSets(kMinusOneSets, i, m_instances.numInstances());
        hashtable = ItemSet.getHashtable(kMinusOneSets, kMinusOneSets.size());
        m_hashtables.addElement(hashtable);
        kSets = ItemSet.pruneItemSets(kSets, hashtable);
        ItemSet.upDateCounters(kSets, m_instances);
        kSets = ItemSet.deleteItemSets(kSets, m_premiseCount,Integer.MAX_VALUE);
        if(kSets.size() == 0)
          return;
        m_Ls.addElement(kSets);
    }
  } 
  
  /** 
   * Method that finds all class association rules.
   *
   * @throws Exception if an attribute is numeric
   */
  private void findCaRulesQuickly() throws Exception {
    
    CaRuleGeneration currentLItemSet;
    // Build rules
    for (int j = 0; j < m_Ls.size(); j++) {
      FastVector currentItemSets = (FastVector)m_Ls.elementAt(j);
      Enumeration enumItemSets = currentItemSets.elements();
      while (enumItemSets.hasMoreElements()) {
        currentLItemSet = new CaRuleGeneration((ItemSet)enumItemSets.nextElement());
        m_best = currentLItemSet.generateRules(m_numRules, m_midPoints,m_priors,m_expectation,
                                        m_instances,m_best,m_count);
        m_count = currentLItemSet.count();
        if(!m_bestChanged && currentLItemSet.change())
                m_bestChanged = true;
        if(m_best.size() >0)
            m_expectation = ((RuleItem)m_best.first()).accuracy();
        else 
            m_expectation =0;
      }
    }
  }


  /**
   * Main method.
   * 
   * @param args the commandline parameters
   */
  public static void main(String[] args) {

    String trainFileString;
    StringBuffer text = new StringBuffer();
    PredictiveApriori apriori = new PredictiveApriori();
    Reader reader;

    try {
      text.append("\n\nPredictiveApriori options:\n\n");
      text.append("-t <training file>\n");
      text.append("\tThe name of the training file.\n");
      Enumeration enu = apriori.listOptions();
      while (enu.hasMoreElements()) {
	Option option = (Option)enu.nextElement();
	text.append(option.synopsis()+'\n');
	text.append(option.description()+'\n');
      }
      trainFileString = Utils.getOption('t', args);
      if (trainFileString.length() == 0) 
	throw new Exception("No training file given!");
      apriori.setOptions(args);
      reader = new BufferedReader(new FileReader(trainFileString));
      if(!apriori.getCar())
        apriori.buildAssociations(new Instances(reader));
      else
        apriori.mineCARs(new Instances(reader));
      System.out.println(apriori);
    } catch(Exception e) {
      e.printStackTrace();
      System.out.println("\n"+e.getMessage()+text);
    }
  }
  
}