winnow.java

MacroWeka扩展了著名数据挖掘工具weka

  }
  
  /**
   * Actual update routine (balanced) for prefiltered instances
   *
   * @exception Exception if something goes wrong
   */
  private void actualUpdateClassifierBalanced(Instance inst) throws Exception {
    
    double posmultiplier,negmultiplier;

    if (!inst.classIsMissing()) {
      double prediction = makePredictionBalanced(inst);
        
      if (prediction != inst.classValue()) {
	m_Mistakes++;
	
	if(prediction == 0) {
	  /* false neg: promote positive, demote negative*/
	  posmultiplier=m_Alpha;
	  negmultiplier=m_Beta;
	} else {
	  /* false pos: demote positive, promote negative */
	  posmultiplier=m_Beta;
	  negmultiplier=m_Alpha;
	}
	int n1 = inst.numValues(); int classIndex = m_Train.classIndex();
	for(int l = 0 ; l < n1 ; l++) {
	  if(inst.index(l) != classIndex && inst.valueSparse(l)==1) {
	    m_predPosVector[inst.index(l)]*=posmultiplier;
	    m_predNegVector[inst.index(l)]*=negmultiplier;
	  }
	}
	//Utils.normalize(m_predPosVector);
	//Utils.normalize(m_predNegVector);
      }
    }
    else {
      System.out.println("CLASS MISSING");
    }
  }

  /**
   * Outputs the prediction for the given instance.
   *
   * @param inst the instance for which prediction is to be computed
   * @return the prediction
   * @exception Exception if something goes wrong
   */
  public double classifyInstance(Instance inst) throws Exception {

    m_ReplaceMissingValues.input(inst);
    m_ReplaceMissingValues.batchFinished();
    Instance filtered = m_ReplaceMissingValues.output();
    m_NominalToBinary.input(filtered);
    m_NominalToBinary.batchFinished();
    filtered = m_NominalToBinary.output();

    if(m_Balanced) {
      return(makePredictionBalanced(filtered));
    } else {
      return(makePrediction(filtered));
    }
  }
  
  /** 
   * Compute the actual prediction for prefiltered instance
   *
   * @param inst the instance for which prediction is to be computed
   * @return the prediction
   * @exception Exception if something goes wrong
   */
  private double makePrediction(Instance inst) throws Exception {

    double total = 0;

    int n1 = inst.numValues(); int classIndex = m_Train.classIndex();
	
    for(int i=0;i<n1;i++) {
      if(inst.index(i) != classIndex && inst.valueSparse(i)==1) {
	total+=m_predPosVector[inst.index(i)];
      }
    }
    
    if(total > m_actualThreshold) {
      return(1);
    } else {
      return(0);
    }
  }
  
  /** 
   * Compute our prediction (Balanced) for prefiltered instance 
   *
   * @param inst the instance for which prediction is to be computed
   * @return the prediction
   * @exception Exception if something goes wrong
   */
  private double makePredictionBalanced(Instance inst) throws Exception {
    double total=0;
	
    int n1 = inst.numValues(); int classIndex = m_Train.classIndex();
    for(int i=0;i<n1;i++) {
      if(inst.index(i) != classIndex && inst.valueSparse(i)==1) {
	total+=(m_predPosVector[inst.index(i)]-m_predNegVector[inst.index(i)]);
      }
    }
     
    if(total > m_actualThreshold) {
      return(1);
    } else {
      return(0);
    }
  }

  /**
   * Returns textual description of the classifier.
   */
  public String toString() {

    if(m_predPosVector==null)
      return("Winnow: No model built yet.");
	   
    String result = "Winnow\n\nAttribute weights\n\n";
	
    int classIndex = m_Train.classIndex();

    if(!m_Balanced) {
      for( int i = 0 ; i < m_Train.numAttributes(); i++) {
	if(i!=classIndex)
	  result += "w" + i + " " + m_predPosVector[i] + "\n";
      }
    } else {
      for( int i = 0 ; i < m_Train.numAttributes(); i++) {
	if(i!=classIndex) {
	  result += "w" + i + " p " + m_predPosVector[i];
	  result += " n " + m_predNegVector[i];
	  
	  double wdiff=m_predPosVector[i]-m_predNegVector[i];
	  
	  result += " d " + wdiff + "\n";
	}
      }
    }
    result += "\nCumulated mistake count: " + m_Mistakes + "\n\n";
	
    return(result);
  }
     
  /**
   * Returns the tip text for this property
   * @return tip text for this property suitable for
   * displaying in the explorer/experimenter gui
   */
  public String balancedTipText() {
    return "Whether to use the balanced version of the algorithm.";
  }

  /**
   * Get the value of Balanced.
   *
   * @return Value of Balanced.
   */
  public boolean getBalanced() {
    
    return m_Balanced;
  }
  
  /**
   * Set the value of Balanced.
   *
   * @param b  Value to assign to Balanced.
   */
  public void setBalanced(boolean b) {
    
    m_Balanced = b;
  }
     
  /**
   * Returns the tip text for this property
   * @return tip text for this property suitable for
   * displaying in the explorer/experimenter gui
   */
  public String alphaTipText() {
    return "Promotion coefficient alpha.";
  }
  
  /**
   * Get the value of Alpha.
   *
   * @return Value of Alpha.
   */
  public double getAlpha() {
    
    return(m_Alpha);
  }
  
  /**
   * Set the value of Alpha.
   *
   * @param a  Value to assign to Alpha.
   */
  public void setAlpha(double a) {
    
    m_Alpha = a;
  }
     
  /**
   * Returns the tip text for this property
   * @return tip text for this property suitable for
   * displaying in the explorer/experimenter gui
   */
  public String betaTipText() {
    return "Demotion coefficient beta.";
  }
  
  /**
   * Get the value of Beta.
   *
   * @return Value of Beta.
   */
  public double getBeta() {
    
    return(m_Beta);
  }
  
  /**
   * Set the value of Beta.
   *
   * @param b  Value to assign to Beta.
   */
  public void setBeta(double b) {
    
    m_Beta = b;
  }
     
  /**
   * Returns the tip text for this property
   * @return tip text for this property suitable for
   * displaying in the explorer/experimenter gui
   */
  public String thresholdTipText() {
    return "Prediction threshold (-1 means: set to number of attributes).";
  }
  
  /**
   * Get the value of Threshold.
   *
   * @return Value of Threshold.
   */
  public double getThreshold() {
    
    return m_Threshold;
  }
  
  /**
   * Set the value of Threshold.
   *
   * @param t  Value to assign to Threshold.
   */
  public void setThreshold(double t) {
    
    m_Threshold = t;
  }
     
  /**
   * Returns the tip text for this property
   * @return tip text for this property suitable for
   * displaying in the explorer/experimenter gui
   */
  public String defaultWeightTipText() {
    return "Initial value of weights/coefficients.";
  }
  
  /**
   * Get the value of defaultWeight.
   *
   * @return Value of defaultWeight.
   */
  public double getDefaultWeight() {
    
    return m_defaultWeight;
  }
  
  /**
   * Set the value of defaultWeight.
   *
   * @param w  Value to assign to defaultWeight.
   */
  public void setDefaultWeight(double w) {
    
    m_defaultWeight = w;
  }
     
  /**
   * Returns the tip text for this property
   * @return tip text for this property suitable for
   * displaying in the explorer/experimenter gui
   */
  public String numIterationsTipText() {
    return "The number of iterations to be performed.";
  }
  
  /**
   * Get the value of numIterations.
   *
   * @return Value of numIterations.
   */
  public int getNumIterations() {
    
    return m_numIterations;
  }
  
  /**
   * Set the value of numIterations.
   *
   * @param v  Value to assign to numIterations.
   */
  public void setNumIterations(int v) {
    
    m_numIterations = v;
  }
     
  /**
   * Returns the tip text for this property
   * @return tip text for this property suitable for
   * displaying in the explorer/experimenter gui
   */
  public String seedTipText() {
    return "Random number seed used for data shuffling (-1 means no "
      + "randomization).";
  }

  /**
   * Get the value of Seed.
   *
   * @return Value of Seed.
   */
  public int getSeed() {
    
    return m_Seed;
  }
  
  /**
   * Set the value of Seed.
   *
   * @param v  Value to assign to Seed.
   */
  public void setSeed(int v) {
    
    m_Seed = v;
  }
  
  /**
   * Main method.
   */
  public static void main(String[] argv) {
    
    try {
      System.out.println(Evaluation.evaluateModel(new Winnow(), argv));
    } catch (Exception e) {
      System.out.println(e.getMessage());
    }
  }
}