Examples of weka.classifiers.trees.J48

weka.classifiers.trees.j48.PruneableClassifierTree

Class for generating a pruned or unpruned C4.5 decision tree. For more information, see

Ross Quinlan (1993). C4.5: Programs for Machine Learning. Morgan Kaufmann Publishers, San Mateo, CA.

BibTeX:

 @book{Quinlan1993, address = {San Mateo, CA}, author = {Ross Quinlan}, publisher = {Morgan Kaufmann Publishers}, title = {C4.5: Programs for Machine Learning}, year = {1993} }

Valid options are:

 -U Use unpruned tree.

 -C <pruning confidence> Set confidence threshold for pruning. (default 0.25)

 -M <minimum number of instances> Set minimum number of instances per leaf. (default 2)

 -R Use reduced error pruning.

 -N <number of folds> Set number of folds for reduced error pruning. One fold is used as pruning set. (default 3)

 -B Use binary splits only.

 -S Don't perform subtree raising.

 -L Do not clean up after the tree has been built.

 -A Laplace smoothing for predicted probabilities.

 -Q <seed> Seed for random data shuffling (default 1).

@author Eibe Frank (eibe@cs.waikato.ac.nz) @version $Revision: 1.9 $

    
    Instances [] localInstances,localPruneInstances;
    int index,ind;
    int i,j;
    double sumOfWeights;
    NoSplit noSplit;
    
    m_train = null;
    m_test = null;
    m_isLeaf = false;
    m_isEmpty = false;
    m_sons = null;
    indeX = 0;
    sumOfWeights = data.sumOfWeights();
    noSplit = new NoSplit (new Distribution((Instances)data));
    if (leaf)
      m_localModel = noSplit;
    else
      m_localModel = m_toSelectModel.selectModel(data);
    if (m_localModel.numSubsets() > 1) {

View Full Code Here

    
    Instances [] localTrain,localTest;
    int index,ind;
    int i,j;
    double sumOfWeights;
    NoSplit noSplit;
    
    m_train = null;
    m_isLeaf = false;
    m_isEmpty = false;
    m_sons = null;
    indeX = 0;
    sumOfWeights = train.sumOfWeights();
    noSplit = new NoSplit (new Distribution((Instances)train));
    if (leaf)
      m_localModel = noSplit;
    else
      m_localModel = m_toSelectModel.selectModel(train, test);
    m_test = new Distribution(test, m_localModel);

View Full Code Here

    errorsTree = errorsForTree();
    errorsLeaf = errorsForLeaf();
    if (Utils.smOrEq(errorsLeaf,errorsTree)){ 
      m_isLeaf = true;
      m_sons = null;
      m_localModel = new NoSplit(localModel().distribution());
    }
  }

View Full Code Here

    
    Instances [] localInstances,localPruneInstances;
    int index,ind;
    int i,j;
    double sumOfWeights;
    NoSplit noSplit;
    
    m_train = null;
    m_test = null;
    m_isLeaf = false;
    m_isEmpty = false;
    m_sons = null;
    indeX = 0;
    sumOfWeights = data.sumOfWeights();
    noSplit = new NoSplit (new Distribution((Instances)data));
    if (leaf)
      m_localModel = noSplit;
    else
      m_localModel = m_toSelectModel.selectModel(data);
    if (m_localModel.numSubsets() > 1) {

View Full Code Here

    errorsTree = getEstimatedErrorsForTree();
    errorsLeaf = getEstimatedErrorsForLeaf();
    if (Utils.smOrEq(errorsLeaf,errorsTree+0.1)) { // +0.1 as in C4.5
      m_isLeaf = true;
      m_sons = null;
      m_localModel = new NoSplit(localModel().distribution());
    }
  }

View Full Code Here

      double[][] dataZs, double[][] dataWs) throws Exception{


    int numAttributes = data.numAttributes();


    if (numAttributes < 2) throw new Exception("Can't select Model without non-class attribute");
    if (data.numInstances() < m_minNumInstances) return new NoSplit(new Distribution(data));




    double bestGain = -Double.MAX_VALUE;
    int bestAttribute = -1;


    //try split on every attribute
    for (int i = 0; i < numAttributes; i++) {
      if (i != data.classIndex()) {


  //build split
  ResidualSplit split = new ResidualSplit(i);      
  split.buildClassifier(data, dataZs, dataWs);


  if (split.checkModel(m_minNumInstances)){


    //evaluate split 
    double gain = split.entropyGain();  
    if (gain > bestGain) {
      bestGain = gain;
      bestAttribute = i;
    }
  }
      }          
    }     


    if (bestGain >= m_minInfoGain){
      //return best split
      ResidualSplit split = new ResidualSplit(bestAttribute);
      split.buildClassifier(data, dataZs, dataWs);  
      return split;      
    } else {      
      //could not find any split with enough information gain
      return new NoSplit(new Distribution(data));      
    }
  }

View Full Code Here

        m_sons = null;
        m_isLeaf = true;
        m_hasConstr=false;
        m_leafclass=m_localModel.distribution().maxClass();
        // Get NoSplit Model for node.
        m_localModel = new NoSplit(m_localModel.distribution());
        treeError=errorsLeaf;


      }else{
        // Decide if Constructor is best choice.
        if (Utils.smOrEq(errorsConstModel, errorsTree)) {
          // Free son Trees
          m_sons = null;
          m_isLeaf = true;
          m_hasConstr =true;
          // Get NoSplit Model for node.
          m_localModel = new NoSplit(m_localModel.distribution());
          treeError=errorsConstModel;
        } else
          treeError=errorsTree;
      }
    }

View Full Code Here

        m_sons = null;
        m_isLeaf = true;
        m_hasConstr=false;
        m_leafclass=m_localModel.distribution().maxClass();
        // Get NoSplit Model for node.
        m_localModel = new NoSplit(m_localModel.distribution());
        treeError=errorsLeaf;


      }else{
        // Decide if Constructor is best choice.
        if (Utils.smOrEq(errorsConstModel, errorsTree)) {
          // Free son Trees
          m_sons = null;
          m_isLeaf = true;
          m_hasConstr =true;
          // Get NoSplit Model for node.
          m_localModel = new NoSplit(m_localModel.distribution());
          treeError=errorsConstModel;
        } else
          treeError=errorsTree;
      }
    }

View Full Code Here

        m_sons = null;
        m_isLeaf = true;
        m_hasConstr=false;
        m_leafclass=m_localModel.distribution().maxClass();
        // Get NoSplit Model for node.
        m_localModel = new NoSplit(m_localModel.distribution());
        treeError=errorsLeaf;


      } else{
        treeError=errorsTree;
      }

View Full Code Here

    
    try {
      if (!m_reducedErrorPruning)
        result = new C45PruneableClassifierTree(null, !m_unpruned, m_CF, m_subtreeRaising, !m_noCleanup).getCapabilities();
      else
        result = new PruneableClassifierTree(null, !m_unpruned, m_numFolds, !m_noCleanup, m_Seed).getCapabilities();
    }
    catch (Exception e) {
      result = new Capabilities(this);
    }

View Full Code Here

0 1 2 3 4 5 6 7 8

TOP

Related Classes of weka.classifiers.trees.J48

de.tudarmstadt.ukp.similarity.experiments.coling2012.util.Evaluator

edu.brown.markov.FeatureClusterer

edu.uga.cs.fluxbuster.classification.Classifier

etc.aloe.cscw2013.WekaModelTest

eu.admire.ogsadai.activity.weka.WekaUtilitiesTest

trust.weka4jason.WekaBinClassifier

trust.weka4jason.WekaClusterClassifier

weka.classifiers.bayes.NaiveBayesUpdateable

weka.classifiers.Evaluation

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.