Examples of weka.core.Instances.sumOfWeights()

weka.core.Instances.sumOfWeights()
Computes the sum of all the instances' weights. @return the sum of all the instances' weights as a double

  
    /* Calculate accuracy parameters for all the antecedents in this rule */
    for(int x=0; x<size; x++){
      Antd antd=(Antd)m_Antds.elementAt(x);
      Instances newData = new Instances(data);
      if(Utils.eq(newData.sumOfWeights(),0.0))
  break;
      
      data = new Instances(newData, newData.numInstances()); // Make data empty
      
      for(int y=0; y<newData.numInstances(); y++){

View Full Code Here

    int numCl = data.numClasses();
    m_Root = new Ridor_node();
    m_Class = instances.classAttribute();     // The original class label
  
    int index = data.classIndex();
    m_Cover = data.sumOfWeights();
  
    /* Create a binary attribute */
    FastVector binary_values = new FastVector(2);
    binary_values.addElement("otherClasses");
    binary_values.addElement("defClass");

View Full Code Here

        }
      
      data.setClassIndex(index);           // Position new class label 
      
      /* Build a set of rules */
      double classCount = data.sumOfWeights() - isPure[i];
      Vector ruleset = new Vector();
      double wAcRt = buildRuleset(data, classCount, ruleset); 
      
      if(Utils.gr(wAcRt, maxAcRt)){
        finalRules = ruleset;

View Full Code Here

     */
    private double buildRuleset(Instances insts, double classCount, Vector ruleset) 
      throws Exception {      
      Instances data = new Instances(insts);
      double wAcRt = 0;  // The weighted accuracy rate of this ruleset
      double total = data.sumOfWeights();
      
      while( classCount >= m_Folds ){      // Data is not pure
  RidorRule bestRule = null;
  double bestWorthRate= -1;        // The best worth achieved by
  double bestWorth = -1;           // randomization of the data

View Full Code Here

  throw new UnsupportedClassTypeException(" Only nominal class, please.");
      if(instances.numClasses() != 2)
  throw new Exception(" Only 2 classes, please.");
      
      Instances data = new Instances(instances);
      if(Utils.eq(data.sumOfWeights(),0))
  throw new Exception(" No training data.");
      
      data.deleteWithMissingClass();
      if(Utils.eq(data.sumOfWeights(),0))
  throw new Exception(" The class labels of all the training data are missing.");

View Full Code Here

      Instances data = new Instances(instances);
      if(Utils.eq(data.sumOfWeights(),0))
  throw new Exception(" No training data.");
      
      data.deleteWithMissingClass();
      if(Utils.eq(data.sumOfWeights(),0))
  throw new Exception(" The class labels of all the training data are missing.");  
      
      if(data.numInstances() < m_Folds)
  throw new Exception(" Not enough data for REP.");

View Full Code Here

      public void grow(Instances data) throws Exception {
      if(m_Consequent == -1)
  throw new Exception(" Consequent not set yet.");
      
      Instances growData = data;           
      double sumOfWeights = growData.sumOfWeights();
      if(!Utils.gr(sumOfWeights, 0.0))
  return;
      
      /* Compute the default accurate rate of the growing data */
      double defAccu = computeDefAccu(growData);

View Full Code Here

     *                 the whole pruning data instead of the data covered
     */    
    public void prune(Instances pruneData, boolean useWhole){
  Instances data = pruneData;
  
      double total = data.sumOfWeights();
      if(!Utils.gr(total, 0.0))
  return;
  
      /* The default accurate # and rate on pruning data */
      double defAccu=computeDefAccu(data);

View Full Code Here

  current.setValue(boostData.classIndex(), z);
  current.setWeight(current.weight() * w);
      }
      
      // Scale the weights (helps with some base learners)
      double sumOfWeights = boostData.sumOfWeights();
      double scalingFactor = (double)origSumOfWeights / sumOfWeights;
      for (int i = 0; i < probs.length; i++) {
  Instance current = boostData.instance(i);
  current.setWeight(current.weight() * scalingFactor);
      }

View Full Code Here

      value = 0;  
      
      // Compute minimum number of Instances required in each split
      double minSplit;
      if(m_ClassAttribute.isNominal()){
  minSplit =  0.1 * (data.sumOfWeights()) /
    ((double)m_ClassAttribute.numValues());
  if (Utils.smOrEq(minSplit,m_MinNo)) 
    minSplit = m_MinNo;
  else if (Utils.gr(minSplit,25)) 
    minSplit = 25;

View Full Code Here

0 1 2 3

TOP

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.