m5base.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

  protected boolean getGenerateRules() {
    return m_generateRules;
  } 

  /**
   * Returns the tip text for this property
   * 
   * @return 		tip text for this property suitable for
   * 			displaying in the explorer/experimenter gui
   */
  public String useUnsmoothedTipText() {
    return "Whether to use unsmoothed predictions.";
  }

  /**
   * Use unsmoothed predictions
   * 
   * @param s true if unsmoothed predictions are to be used
   */
  public void setUseUnsmoothed(boolean s) {
    m_unsmoothedPredictions = s;
  } 

  /**
   * Get whether or not smoothing is being used
   * 
   * @return true if unsmoothed predictions are to be used
   */
  public boolean getUseUnsmoothed() {
    return m_unsmoothedPredictions;
  } 

  /**
   * Returns the tip text for this property
   * 
   * @return 		tip text for this property suitable for
   * 			displaying in the explorer/experimenter gui
   */
  public String buildRegressionTreeTipText() {
    return "Whether to generate a regression tree/rule instead of a model tree/rule.";
  }

  /**
   * Get the value of regressionTree.
   *
   * @return Value of regressionTree.
   */
  public boolean getBuildRegressionTree() {
    
    return m_regressionTree;
  }
  
  /**
   * Set the value of regressionTree.
   *
   * @param newregressionTree Value to assign to regressionTree.
   */
  public void setBuildRegressionTree(boolean newregressionTree) {
    
    m_regressionTree = newregressionTree;
  }

  /**
   * Returns the tip text for this property
   * 
   * @return 		tip text for this property suitable for
   * 			displaying in the explorer/experimenter gui
   */
  public String minNumInstancesTipText() {
    return "The minimum number of instances to allow at a leaf node.";
  }

  /**
   * Set the minimum number of instances to allow at a leaf node
   *
   * @param minNum the minimum number of instances
   */
  public void setMinNumInstances(double minNum) {
    m_minNumInstances = minNum;
  }

  /**
   * Get the minimum number of instances to allow at a leaf node
   *
   * @return a <code>double</code> value
   */
  public double getMinNumInstances() {
    return m_minNumInstances;
  }

  /**
   * Returns default capabilities of the classifier, i.e., of LinearRegression.
   *
   * @return      the capabilities of this classifier
   */
  public Capabilities getCapabilities() {
    return new LinearRegression().getCapabilities();
  }

  /**
   * Generates the classifier.
   * 
   * @param data set of instances serving as training data
   * @throws Exception if the classifier has not been generated
   * successfully
   */
  public void buildClassifier(Instances data) throws Exception {
    // can classifier handle the data?
    getCapabilities().testWithFail(data);

    // remove instances with missing class
    data = new Instances(data);
    data.deleteWithMissingClass();
    
    m_instances = new Instances(data);

    m_replaceMissing = new ReplaceMissingValues();
    m_replaceMissing.setInputFormat(m_instances);
    m_instances = Filter.useFilter(m_instances, m_replaceMissing);

    m_nominalToBinary = new NominalToBinary();
    m_nominalToBinary.setInputFormat(m_instances);
    m_instances = Filter.useFilter(m_instances, m_nominalToBinary);

    m_removeUseless = new RemoveUseless();
    m_removeUseless.setInputFormat(m_instances);
    m_instances = Filter.useFilter(m_instances, m_removeUseless);
    
    m_instances.randomize(new Random(1));

    m_ruleSet = new FastVector();

    Rule tempRule;

    if (m_generateRules) {
      Instances tempInst = m_instances;
     
      do {
	tempRule = new Rule();
	tempRule.setSmoothing(!m_unsmoothedPredictions);
	tempRule.setRegressionTree(m_regressionTree);
	tempRule.setUnpruned(m_useUnpruned);
	tempRule.setSaveInstances(false);
	tempRule.setMinNumInstances(m_minNumInstances);
	tempRule.buildClassifier(tempInst);
	m_ruleSet.addElement(tempRule);
	//	System.err.println("Built rule : "+tempRule.toString());
	tempInst = tempRule.notCoveredInstances();
      } while (tempInst.numInstances() > 0);
    } else {
      // just build a single tree
      tempRule = new Rule();

      tempRule.setUseTree(true);
      //      tempRule.setGrowFullTree(true);
      tempRule.setSmoothing(!m_unsmoothedPredictions);
      tempRule.setSaveInstances(m_saveInstances);
      tempRule.setRegressionTree(m_regressionTree);
      tempRule.setUnpruned(m_useUnpruned);
      tempRule.setMinNumInstances(m_minNumInstances);

      Instances temp_train;

      temp_train = m_instances;

      tempRule.buildClassifier(temp_train);

      m_ruleSet.addElement(tempRule);      

      // save space
      m_instances = new Instances(m_instances, 0);
      //      System.err.print(tempRule.m_topOfTree.treeToString(0));
    } 
  } 

  /**
   * Calculates a prediction for an instance using a set of rules
   * or an M5 model tree
   * 
   * @param inst the instance whos class value is to be predicted
   * @return the prediction
   * @throws Exception if a prediction can't be made.
   */
  public double classifyInstance(Instance inst) throws Exception {
    Rule   temp;
    double prediction = 0;
    boolean success = false;

    m_replaceMissing.input(inst);
    inst = m_replaceMissing.output();
    m_nominalToBinary.input(inst);
    inst = m_nominalToBinary.output();
    m_removeUseless.input(inst);
    inst = m_removeUseless.output();

    if (m_ruleSet == null) {
      throw new Exception("Classifier has not been built yet!");
    } 

    if (!m_generateRules) {
      temp = (Rule) m_ruleSet.elementAt(0);
      return temp.classifyInstance(inst);
    } 

    boolean cont;
    int     i;

    for (i = 0; i < m_ruleSet.size(); i++) {
      cont = false;
      temp = (Rule) m_ruleSet.elementAt(i);

      try {
	prediction = temp.classifyInstance(inst);
	success = true;
      } catch (Exception e) {
	cont = true;
      } 

      if (!cont) {
	break;
      } 
    } 

    if (!success) {
      System.out.println("Error in predicting (DecList)");
    } 
    return prediction;
  } 

  /**
   * Returns a description of the classifier
   * 
   * @return a description of the classifier as a String
   */
  public String toString() {
    StringBuffer text = new StringBuffer();
    Rule	 temp;

    if (m_ruleSet == null) {
      return "Classifier hasn't been built yet!";
    } 

    if (m_generateRules) {
      text.append("M5 "
		  + ((m_useUnpruned == true)
		     ? "unpruned "
		     : "pruned ")
		  + ((m_regressionTree == true) 
		     ?  "regression "
		     : "model ")
		  + "rules ");

      if (!m_unsmoothedPredictions) {
	text.append("\n(using smoothed linear models) ");
      }

      text.append(":\n");

      text.append("Number of Rules : " + m_ruleSet.size() + "\n\n");

      for (int j = 0; j < m_ruleSet.size(); j++) {
	temp = (Rule) m_ruleSet.elementAt(j);

	text.append("Rule: " + (j + 1) + "\n");
	text.append(temp.toString());
      } 
    } else {
      temp = (Rule) m_ruleSet.elementAt(0);
      text.append(temp.toString());
    } 
    return text.toString();
  } 

  /**
   * Returns an enumeration of the additional measure names
   * @return an enumeration of the measure names
   */
  public Enumeration enumerateMeasures() {
    Vector newVector = new Vector(1);
    newVector.addElement("measureNumRules");
    return newVector.elements();
  }

  /**
   * Returns the value of the named measure
   * @param additionalMeasureName the name of the measure to query for its value
   * @return the value of the named measure
   * @throws Exception if the named measure is not supported
   */
  public double getMeasure(String additionalMeasureName) 
    {
    if (additionalMeasureName.compareToIgnoreCase("measureNumRules") == 0) {
      return measureNumRules();
    } else {
      throw new IllegalArgumentException(additionalMeasureName 
					 + " not supported (M5)");
    }
  }

  /**
   * return the number of rules
   * @return the number of rules (same as # linear models &
   * # leaves in the tree)
   */
  public double measureNumRules() {
    if (m_generateRules) {
      return m_ruleSet.size();
    }
    return ((Rule)m_ruleSet.elementAt(0)).m_topOfTree.numberOfLinearModels();
  }

  public RuleNode getM5RootNode() {
    Rule temp = (Rule) m_ruleSet.elementAt(0);
    return temp.getM5RootNode();
  }
}