Examples of weka.filters.unsupervised.instance.RemoveWithValues

Package weka.filters.unsupervised.instance

Examples of weka.filters.unsupervised.instance.RemoveWithValues

weka.filters.unsupervised.instance.RemoveWithValues

Filters instances according to the value of an attribute.

Valid options are:

 -C <num> Choose attribute to be used for selection.

 -S <num> Numeric value to be used for selection on numeric attribute. Instances with values smaller than given value will be selected. (default 0)

 -L <index1,index2-index4,...> Range of label indices to be used for selection on nominal attribute. First and last are valid indexes. (default all values)

 -M Missing values count as a match. This setting is independent of the -V option. (default missing values don't match)

 -V Invert matching sense.

 -H When selecting on nominal attributes, removes header references to excluded values.

@author Eibe Frank (eibe@cs.waikato.ac.nz) @version $Revision: 6798 $

      m_ClassFilters = new Filter[numClassifiers];
      m_SumOfWeights = new double[numClassifiers];


      // generate the classifiers
      for (int i=0; i<numClassifiers; i++) {
  RemoveWithValues classFilter = new RemoveWithValues();
  classFilter.setAttributeIndex("" + (insts.classIndex() + 1));
  classFilter.setModifyHeader(true);
  classFilter.setInvertSelection(true);
  classFilter.setNominalIndicesArr((int[])pairs.elementAt(i));
  Instances tempInstances = new Instances(insts, 0);
  tempInstances.setClassIndex(-1);
  classFilter.setInputFormat(tempInstances);
  newInsts = Filter.useFilter(insts, classFilter);
  if (newInsts.numInstances() > 0) {
    newInsts.setClassIndex(insts.classIndex());
    m_Classifiers[i].buildClassifier(newInsts);
    m_ClassFilters[i] = classFilter;
          m_SumOfWeights[i] = newInsts.sumOfWeights();
  } else {
    m_Classifiers[i] = null;
    m_ClassFilters[i] = null;
  }
      }


      // construct a two-class header version of the dataset
      m_TwoClassDataset = new Instances(insts, 0);
      int classIndex = m_TwoClassDataset.classIndex();
      m_TwoClassDataset.setClassIndex(-1);
      m_TwoClassDataset.deleteAttributeAt(classIndex);
      FastVector classLabels = new FastVector();
      classLabels.addElement("class0");
      classLabels.addElement("class1");
      m_TwoClassDataset.insertAttributeAt(new Attribute("class", classLabels),
            classIndex);
      m_TwoClassDataset.setClassIndex(classIndex);


    } else { // use error correcting code style methods
      Code code = null;
      switch (m_Method) {
      case METHOD_ERROR_EXHAUSTIVE:
        code = new ExhaustiveCode(numClassifiers);
        break;
      case METHOD_ERROR_RANDOM:
        code = new RandomCode(numClassifiers, 
                              (int)(numClassifiers * m_RandomWidthFactor),
            insts);
        break;
      case METHOD_1_AGAINST_ALL:
        code = new StandardCode(numClassifiers);
        break;
      default:
        throw new Exception("Unrecognized correction code type");
      }
      numClassifiers = code.size();
      m_Classifiers = AbstractClassifier.makeCopies(m_Classifier, numClassifiers);
      m_ClassFilters = new MakeIndicator[numClassifiers];
      for (int i = 0; i < m_Classifiers.length; i++) {
  m_ClassFilters[i] = new MakeIndicator();
  MakeIndicator classFilter = (MakeIndicator) m_ClassFilters[i];
  classFilter.setAttributeIndex("" + (insts.classIndex() + 1));
  classFilter.setValueIndices(code.getIndices(i));
  classFilter.setNumeric(false);
  classFilter.setInputFormat(insts);
  newInsts = Filter.useFilter(insts, m_ClassFilters[i]);
  m_Classifiers[i].buildClassifier(newInsts);
      }
    }
    m_ClassAttribute = insts.classAttribute();

View Full Code Here

      m_ClassFilters = new Filter[numClassifiers];
      m_SumOfWeights = new double[numClassifiers];


      // generate the classifiers
      for (int i=0; i<numClassifiers; i++) {
  RemoveWithValues classFilter = new RemoveWithValues();
  classFilter.setAttributeIndex("" + (insts.classIndex() + 1));
  classFilter.setModifyHeader(true);
  classFilter.setInvertSelection(true);
  classFilter.setNominalIndicesArr((int[])pairs.elementAt(i));
  Instances tempInstances = new Instances(insts, 0);
  tempInstances.setClassIndex(-1);
  classFilter.setInputFormat(tempInstances);
  newInsts = Filter.useFilter(insts, classFilter);
  if (newInsts.numInstances() > 0) {
    newInsts.setClassIndex(insts.classIndex());
    m_Classifiers[i].buildClassifier(newInsts);
    m_ClassFilters[i] = classFilter;
          m_SumOfWeights[i] = newInsts.sumOfWeights();
  } else {
    m_Classifiers[i] = null;
    m_ClassFilters[i] = null;
  }
      }


      // construct a two-class header version of the dataset
      m_TwoClassDataset = new Instances(insts, 0);
      int classIndex = m_TwoClassDataset.classIndex();
      m_TwoClassDataset.setClassIndex(-1);
      m_TwoClassDataset.deleteAttributeAt(classIndex);
      FastVector classLabels = new FastVector();
      classLabels.addElement("class0");
      classLabels.addElement("class1");
      m_TwoClassDataset.insertAttributeAt(new Attribute("class", classLabels),
            classIndex);
      m_TwoClassDataset.setClassIndex(classIndex);


    } else { // use error correcting code style methods
      Code code = null;
      switch (m_Method) {
      case METHOD_ERROR_EXHAUSTIVE:
        code = new ExhaustiveCode(numClassifiers);
        break;
      case METHOD_ERROR_RANDOM:
        code = new RandomCode(numClassifiers, 
                              (int)(numClassifiers * m_RandomWidthFactor),
            insts);
        break;
      case METHOD_1_AGAINST_ALL:
        code = new StandardCode(numClassifiers);
        break;
      default:
        throw new Exception("Unrecognized correction code type");
      }
      numClassifiers = code.size();
      m_Classifiers = Classifier.makeCopies(m_Classifier, numClassifiers);
      m_ClassFilters = new MakeIndicator[numClassifiers];
      for (int i = 0; i < m_Classifiers.length; i++) {
  m_ClassFilters[i] = new MakeIndicator();
  MakeIndicator classFilter = (MakeIndicator) m_ClassFilters[i];
  classFilter.setAttributeIndex("" + (insts.classIndex() + 1));
  classFilter.setValueIndices(code.getIndices(i));
  classFilter.setNumeric(false);
  classFilter.setInputFormat(insts);
  newInsts = Filter.useFilter(insts, m_ClassFilters[i]);
  m_Classifiers[i].buildClassifier(newInsts);
      }
    }
    m_ClassAttribute = insts.classAttribute();

View Full Code Here

    // Create two successors if necessary
    m_FirstSuccessor = new DataNearBalancedND();
    if (first == 1) {
      m_FirstSuccessor.m_Range = m_Range;
    } else {
      RemoveWithValues rwv = new RemoveWithValues();
      rwv.setInvertSelection(true);
      rwv.setNominalIndices(m_Range.getRanges());
      rwv.setAttributeIndex("" + (data.classIndex() + 1));
      rwv.setInputFormat(data);
      Instances firstSubset = Filter.useFilter(data, rwv);
      m_FirstSuccessor.generateClassifierForNode(firstSubset, m_Range, 
                                                 rand, classifier, m_classifiers,
                                                 instsNumAllClasses);
    }
    m_SecondSuccessor = new DataNearBalancedND();
    if (second == 1) {
      m_SecondSuccessor.m_Range = secondRange;
    } else {
      RemoveWithValues rwv = new RemoveWithValues();
      rwv.setInvertSelection(true);
      rwv.setNominalIndices(secondRange.getRanges());
      rwv.setAttributeIndex("" + (data.classIndex() + 1));
      rwv.setInputFormat(data);
      Instances secondSubset = Filter.useFilter(data, rwv);
      m_SecondSuccessor = new DataNearBalancedND();
      
      m_SecondSuccessor.generateClassifierForNode(secondSubset, secondRange, 
                                                  rand, classifier, m_classifiers,

View Full Code Here

    // Create two successors if necessary
    m_FirstSuccessor = new ClassBalancedND();
    if (first == 1) {
      m_FirstSuccessor.m_Range = m_Range;
    } else {
      RemoveWithValues rwv = new RemoveWithValues();
      rwv.setInvertSelection(true);
      rwv.setNominalIndices(m_Range.getRanges());
      rwv.setAttributeIndex("" + (data.classIndex() + 1));
      rwv.setInputFormat(data);
      Instances firstSubset = Filter.useFilter(data, rwv);
      m_FirstSuccessor.generateClassifierForNode(firstSubset, m_Range, 
                                                 rand, classifier, m_classifiers);
    }
    m_SecondSuccessor = new ClassBalancedND();
    if (second == 1) {
      m_SecondSuccessor.m_Range = secondRange;
    } else {
      RemoveWithValues rwv = new RemoveWithValues();
      rwv.setInvertSelection(true);
      rwv.setNominalIndices(secondRange.getRanges());
      rwv.setAttributeIndex("" + (data.classIndex() + 1));
      rwv.setInputFormat(data);
      Instances secondSubset = Filter.useFilter(data, rwv);
      m_SecondSuccessor = new ClassBalancedND();
      
      m_SecondSuccessor.generateClassifierForNode(secondSubset, secondRange, 
                                                  rand, classifier, m_classifiers);

View Full Code Here

               node.m_right.getString());
      }
      
      // Generate successors
      if (node.m_left.m_left != null) {
        RemoveWithValues rwv = new RemoveWithValues();
        rwv.setInvertSelection(true);
        rwv.setNominalIndices(node.m_left.getString());
        rwv.setAttributeIndex("" + (data.classIndex() + 1));
        rwv.setInputFormat(data);
        Instances firstSubset = Filter.useFilter(data, rwv);
        buildClassifierForNode(node.m_left, firstSubset);
      }
      if (node.m_right.m_left != null) {
        RemoveWithValues rwv = new RemoveWithValues();
        rwv.setInvertSelection(true);
        rwv.setNominalIndices(node.m_right.getString());
        rwv.setAttributeIndex("" + (data.classIndex() + 1));
        rwv.setInputFormat(data);
        Instances secondSubset = Filter.useFilter(data, rwv);
        buildClassifierForNode(node.m_right, secondSubset);
      }
    }
  }

View Full Code Here

      m_ClassFilters = new Filter[numClassifiers];
      m_SumOfWeights = new double[numClassifiers];


      // generate the classifiers
      for (int i=0; i<numClassifiers; i++) {
  RemoveWithValues classFilter = new RemoveWithValues();
  classFilter.setAttributeIndex("" + (insts.classIndex() + 1));
  classFilter.setModifyHeader(true);
  classFilter.setInvertSelection(true);
  classFilter.setNominalIndicesArr((int[])pairs.elementAt(i));
  Instances tempInstances = new Instances(insts, 0);
  tempInstances.setClassIndex(-1);
  classFilter.setInputFormat(tempInstances);
  newInsts = Filter.useFilter(insts, classFilter);
  if (newInsts.numInstances() > 0 || zeroTrainingInstances) {
    newInsts.setClassIndex(insts.classIndex());
    m_Classifiers[i].buildClassifier(newInsts);
    m_ClassFilters[i] = classFilter;
          m_SumOfWeights[i] = newInsts.sumOfWeights();
  } else {
    m_Classifiers[i] = null;
    m_ClassFilters[i] = null;
  }
      }


      // construct a two-class header version of the dataset
      m_TwoClassDataset = new Instances(insts, 0);
      int classIndex = m_TwoClassDataset.classIndex();
      m_TwoClassDataset.setClassIndex(-1);
      m_TwoClassDataset.deleteAttributeAt(classIndex);
      FastVector classLabels = new FastVector();
      classLabels.addElement("class0");
      classLabels.addElement("class1");
      m_TwoClassDataset.insertAttributeAt(new Attribute("class", classLabels),
            classIndex);
      m_TwoClassDataset.setClassIndex(classIndex);


    } else { // use error correcting code style methods
      Code code = null;
      switch (m_Method) {
      case METHOD_ERROR_EXHAUSTIVE:
        code = new ExhaustiveCode(numClassifiers);
        break;
      case METHOD_ERROR_RANDOM:
        code = new RandomCode(numClassifiers, 
                              (int)(numClassifiers * m_RandomWidthFactor),
            insts);
        break;
      case METHOD_1_AGAINST_ALL:
        code = new StandardCode(numClassifiers);
        break;
      default:
        throw new Exception("Unrecognized correction code type");
      }
      numClassifiers = code.size();
      m_Classifiers = AbstractClassifier.makeCopies(m_Classifier, numClassifiers);
      m_ClassFilters = new MakeIndicator[numClassifiers];
      for (int i = 0; i < m_Classifiers.length; i++) {
  m_ClassFilters[i] = new MakeIndicator();
  MakeIndicator classFilter = (MakeIndicator) m_ClassFilters[i];
  classFilter.setAttributeIndex("" + (insts.classIndex() + 1));
  classFilter.setValueIndices(code.getIndices(i));
  classFilter.setNumeric(false);
  classFilter.setInputFormat(insts);
  newInsts = Filter.useFilter(insts, m_ClassFilters[i]);
  m_Classifiers[i].buildClassifier(newInsts);
      }
    }
    m_ClassAttribute = insts.classAttribute();

View Full Code Here

            addFilter.setInputFormat(basicInstances);
        } catch (Exception e) {
            assertTrue(e.getMessage(), false);
        }


        RemoveWithValues removeFilter = new RemoveWithValues();
        removeFilter.setAttributeIndex("3"); //the label attribute
        removeFilter.setNominalIndicesArr(new int[]{0}); //false
        try {
            removeFilter.setInputFormat(addFilter.getOutputFormat());
        } catch (Exception e) {
            assertTrue(e.getMessage(), false);
        }


        FeatureSpecification spec = new FeatureSpecification();

View Full Code Here

     * Returns a new example set containing only those examples with labels.
     *
     * @return
     */
    public ExampleSet onlyLabeled() {
        RemoveWithValues filter = new RemoveWithValues();
        filter.setAttributeIndex("" + (instances.classIndex() + 1));
        filter.setMatchMissingValues(true);
        filter.setInvertSelection(true);


        try {
            filter.setInputFormat(instances);
            Instances result = Filter.useFilter(instances, filter);
            ExampleSet resultSet = new ExampleSet(result);
            return resultSet;
        } catch (Exception ex) {
            System.err.println("Unable to apply filter!");

View Full Code Here

TOP

Related Classes of weka.filters.unsupervised.instance.RemoveWithValues

etc.aloe.cscw2013.FeatureExtractionImplTest

etc.aloe.data.ExampleSet

weka.classifiers.meta.MultiClassClassifier

weka.classifiers.meta.nestedDichotomies.ClassBalancedND

weka.classifiers.meta.nestedDichotomies.DataNearBalancedND

weka.classifiers.meta.nestedDichotomies.ND

weka.core.Attribute

weka.core.Capabilities

weka.core.FastVector

weka.core.Instance

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.