Examples of weka.core.SparseInstance

weka.core.SparseInstance
Class for storing an instance as a sparse vector. A sparse instance only requires storage for those attribute values that are non-zero. Since the objective is to reduce storage requirements for datasets with large numbers of default values, this also includes nominal attributes -- the first nominal value (i.e. that which has index 0) will not require explicit storage, so rearrange your nominal attribute value orderings if necessary. Missing values will be stored explicitly. @author Eibe Frank @version $Revision: 5970 $

        vals[i] = dest.value(i);
      }
    }
    Instance inst = null;
    if (dest instanceof SparseInstance) {
      inst = new SparseInstance(dest.weight(), vals);
    } else {
      inst = new Instance(dest.weight(), vals);
    }
    inst.setDataset(dest.dataset());
    return inst;

View Full Code Here

    indices[num] = instance.index(j);
    num++;
  }
      } 
      if (num == instance.numValues()) {
  inst = new SparseInstance(instance.weight(), vals, indices,
                                  instance.numAttributes());
      } else {
  double []tempVals = new double[num];
  int []tempInd = new int[num];
  System.arraycopy(vals, 0, tempVals, 0, num);
  System.arraycopy(indices, 0, tempInd, 0, num);
  inst = new SparseInstance(instance.weight(), tempVals, tempInd,
                                  instance.numAttributes());
      }
    } else {
      double []vals = new double[getInputFormat().numAttributes()];
      for (int j = 0; j < instance.numAttributes(); j++) {

View Full Code Here

  }
      }
    }
    Instance inst = null;
    if (instance instanceof SparseInstance) {
      inst = new SparseInstance(instance.weight(), vals);
    } else {
      inst = new Instance(instance.weight(), vals);
    }
    inst.setDataset(instance.dataset());
    push(inst);

View Full Code Here

      }  
      double[] tempVals = new double[ind];
      int[] tempInd = new int[ind];
      System.arraycopy(newVals, 0, tempVals, 0, ind);
      System.arraycopy(newIndices, 0, tempInd, 0, ind);
      inst = new SparseInstance(instance.weight(), tempVals, tempInd,
                                instance.numAttributes());
    } else {
      double[] vals = instance.toDoubleArray();
      for (int j = 0; j < getInputFormat().numAttributes(); j++) {
  if (instance.attribute(j).isNumeric() &&

View Full Code Here

      }
    }


    Instance inst = null;
    if (instance instanceof SparseInstance) {
      inst = new SparseInstance(instance.weight(), vals);
    } else {
      inst = new Instance(instance.weight(), vals);
    }
    inst.setDataset(getOutputFormat());
    copyValues(inst, false, instance.dataset(), getOutputFormat());

View Full Code Here

      int current = m_SelectedAttributes[i];
      vals[i + j] = instance.value(current);
    }
    Instance inst = null;
    if (instance instanceof SparseInstance) {
      inst = new SparseInstance(instance.weight(), vals);
    } else {
      inst = new Instance(instance.weight(), vals);
    }


    inst.setDataset(getOutputFormat());

View Full Code Here

    }
    if (m_NewBatch) {
      resetQueue();
      m_NewBatch = false;
    }
    Instance inst = new SparseInstance(instance);
    inst.setDataset(instance.dataset());
    push(inst);
    return true;
  }

View Full Code Here

        newValues[m_outputNumAttributes - 1] = currentInstance.classValue();
      }
      //create new instance with recorded values and add to output dataset
      Instance newInstance;
      if (currentInstance instanceof SparseInstance) {
        newInstance = new SparseInstance(currentInstance.weight(), newValues);
      } else {
        newInstance = new Instance(currentInstance.weight(), newValues);
      }
      output.add(newInstance);
    }

View Full Code Here

      newValues[i] = instanceVector.get(0, i);
    }
    
    // return newly transformed instance
    if (instance instanceof SparseInstance) {
      return new SparseInstance(instance.weight(), newValues);
    } else {
      return new Instance(instance.weight(), newValues);
    }
  }

View Full Code Here

          return;
        }
        
        if((m_classIndex >= 0) &&
        (m_data.attribute(m_classIndex).isNominal())) {
          SparseInstance histClassCounts[];
          histClassCounts = new SparseInstance[m_data.attribute(m_attribIndex).numValues()];
          /*int histClassCounts[][];
          histClassCounts=new int[m_data.attribute(m_attribIndex).numValues()]
                                 [m_data.attribute(m_classIndex).numValues()+1]; */
          
          if (m_as.nominalCounts.length > 0) {
            m_maxValue = m_as.nominalCounts[0];
            for(int i=0; i<m_data.attribute(m_attribIndex).numValues(); i++) {
              if(m_as.nominalCounts[i]>m_maxValue)
          m_maxValue = m_as.nominalCounts[i];
            }
          }
          else {
            m_maxValue = 0;
          }
          
          if(m_colorList.size()==0)
            m_colorList.addElement(Color.black);
          for(int i=m_colorList.size();
          i < m_data.attribute(m_classIndex).numValues()+1; i++) {
            Color pc = m_defaultColors[(i-1) % 10];
            int ija =  (i-1) / 10;
            ija *= 2;
            
            for (int j=0;j<ija;j++) {
              pc = pc.darker();
            }
            
            m_colorList.addElement(pc);
          }
          
          // first sort data on attribute values
          m_data.sort(m_attribIndex);
          double[] tempClassCounts = null;
          int tempAttValueIndex = -1;
          
          for(int k=0; k<m_data.numInstances(); k++) {
            //System.out.println("attrib: "+
            //                   m_data.instance(k).value(m_attribIndex)+
            //                   " class: "+
            //                   m_data.instance(k).value(m_classIndex));
            if(!m_data.instance(k).isMissing(m_attribIndex)) {
              // check to see if we need to allocate some space here
              if (m_data.instance(k).value(m_attribIndex) != tempAttValueIndex) {
                if (tempClassCounts != null) {
                  // set up the sparse instance for the previous bar (if any)
                  int numNonZero = 0;
                  for (int z = 0; z < tempClassCounts.length; z++) {
                    if (tempClassCounts[z] > 0) {
                      numNonZero++;
                    }
                  }
                  double[] nonZeroVals = new double[numNonZero];
                  int[] nonZeroIndices = new int[numNonZero];
                  int count = 0;
                  for (int z = 0; z < tempClassCounts.length; z++) {
                    if (tempClassCounts[z] > 0) {
                      nonZeroVals[count] = tempClassCounts[z];
                      nonZeroIndices[count++] = z;
                    }
                  }
                  SparseInstance tempS = 
                    new SparseInstance(1.0, nonZeroVals, nonZeroIndices, tempClassCounts.length);
                  histClassCounts[tempAttValueIndex] = tempS;
                }
                
                tempClassCounts = new double[m_data.attribute(m_classIndex).numValues() + 1];
                tempAttValueIndex = (int)m_data.instance(k).value(m_attribIndex);
                
                /* histClassCounts[(int)m_data.instance(k).value(m_attribIndex)] = 
                  new double[m_data.attribute(m_classIndex).numValues()+1]; */ 
              }
              if(m_data.instance(k).isMissing(m_classIndex)) {
                /* histClassCounts[(int)m_data.instance(k).value(m_attribIndex)]
                               [0] += m_data.instance(k).weight(); */
                tempClassCounts[0] += m_data.instance(k).weight();
              } else {
                tempClassCounts[(int)m_data.instance(k).value(m_classIndex)+1] 
                                += m_data.instance(k).weight();
                
                /*histClassCounts[(int)m_data.instance(k).value(m_attribIndex)]
                              [(int)m_data.instance(k).value(m_classIndex)+1] += m_data.instance(k).weight();*/
              }
            }
          }
          
          // set up sparse instance for last bar?
          if (tempClassCounts != null) {
            // set up the sparse instance for the previous bar (if any)
            int numNonZero = 0;
            for (int z = 0; z < tempClassCounts.length; z++) {
              if (tempClassCounts[z] > 0) {
                numNonZero++;
              }
            }
            double[] nonZeroVals = new double[numNonZero];
            int[] nonZeroIndices = new int[numNonZero];
            int count = 0;
            for (int z = 0; z < tempClassCounts.length; z++) {
              if (tempClassCounts[z] > 0) {
                nonZeroVals[count] = tempClassCounts[z];
                nonZeroIndices[count++] = z;
              }
            }
            SparseInstance tempS = 
              new SparseInstance(1.0, nonZeroVals, nonZeroIndices, tempClassCounts.length);
            histClassCounts[tempAttValueIndex] = tempS;
          }
          
          /*for(int k=0; k<m_data.numInstances(); k++) {
            //System.out.println("attrib: "+

View Full Code Here

0 1 2 3 4 5 6 7 8 9

TOP

Related Classes of weka.core.SparseInstance

com.jgaap.generics.WEKAAnalysisDriver

distributedRedditAnalyser.bolt.DistributedOzaBoostBolt

distributedRedditAnalyser.bolt.OzaBoostBolt

etc.aloe.filters.WordFeaturesExtractor

org.apache.ctakes.ytex.weka.WekaBagOfWordsExporterImpl

org.apache.ctakes.ytex.weka.WekaFormatterFactory$WekaFormatter

weka.attributeSelection.LatentSemanticAnalysis

weka.attributeSelection.PrincipalComponents

weka.classifiers.timeseries.core.TimeSeriesTranslate

weka.core.converters.ArffLoader$ArffReader

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.