VTK
dox/Infovis/vtkKMeansStatistics.h
Go to the documentation of this file.
00001 /*=========================================================================
00002 
00003 Program:   Visualization Toolkit
00004 Module:    vtkKMeansStatistics.h
00005 
00006 Copyright (c) Ken Martin, Will Schroeder, Bill Lorensen
00007 All rights reserved.
00008 See Copyright.txt or http://www.kitware.com/Copyright.htm for details.
00009 
00010 This software is distributed WITHOUT ANY WARRANTY; without even
00011 the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
00012 PURPOSE.  See the above copyright notice for more information.
00013 
00014 =========================================================================*/
00015 /*-------------------------------------------------------------------------
00016   Copyright 2010 Sandia Corporation.
00017   Under the terms of Contract DE-AC04-94AL85000 with Sandia Corporation,
00018   the U.S. Government retains certain rights in this software.
00019   -------------------------------------------------------------------------*/
00100 #ifndef __vtkKMeansStatistics_h
00101 #define __vtkKMeansStatistics_h
00102 
00103 #include "vtkStatisticsAlgorithm.h"
00104 
00105 class vtkIdTypeArray;
00106 class vtkIntArray;
00107 class vtkDoubleArray;
00108 class vtkKMeansDistanceFunctor;
00109 class vtkMultiBlockDataSet;
00110 
00111 class VTK_INFOVIS_EXPORT vtkKMeansStatistics : public vtkStatisticsAlgorithm
00112 {
00113 public:
00114   vtkTypeMacro(vtkKMeansStatistics, vtkStatisticsAlgorithm);
00115   virtual void PrintSelf( ostream& os, vtkIndent indent );
00116   static vtkKMeansStatistics* New();
00117  
00119 
00120   virtual void SetDistanceFunctor( vtkKMeansDistanceFunctor* );
00121   vtkGetObjectMacro(DistanceFunctor,vtkKMeansDistanceFunctor);
00123 
00125 
00127   vtkSetMacro(DefaultNumberOfClusters, int);
00128   vtkGetMacro(DefaultNumberOfClusters, int);
00130 
00132 
00133   vtkSetStringMacro(KValuesArrayName);
00134   vtkGetStringMacro(KValuesArrayName);
00136 
00138 
00140   vtkSetMacro( MaxNumIterations, int );
00141   vtkGetMacro( MaxNumIterations, int );
00143 
00145 
00147   vtkSetMacro( Tolerance, double );
00148   vtkGetMacro( Tolerance, double );
00150 
00152 
00154   virtual void Aggregate( vtkDataObjectCollection*,
00155                           vtkMultiBlockDataSet* ) { return; };
00157 
00158   //BTX
00160 
00161   virtual bool SetParameter(
00162     const char* parameter, int index, vtkVariant value );
00163   //ETX
00165 
00166 protected:
00167   vtkKMeansStatistics();
00168   ~vtkKMeansStatistics();
00169 
00171 
00172   virtual void Learn( vtkTable* inData,
00173                       vtkTable* inParameters,
00174                       vtkMultiBlockDataSet* outMeta );
00176 
00178   virtual void Derive( vtkMultiBlockDataSet* );
00179 
00181 
00182   virtual void Assess( vtkTable*, 
00183                        vtkMultiBlockDataSet*, 
00184                        vtkTable* );
00186 
00188 
00189   virtual void Test( vtkTable*,
00190                      vtkMultiBlockDataSet*,
00191                      vtkTable* ) { return; };
00193 
00194   //BTX  
00196 
00197   virtual void SelectAssessFunctor( vtkTable* inData, 
00198                                     vtkDataObject* inMeta,
00199                                     vtkStringArray* rowNames,
00200                                     AssessFunctor*& dfunc );
00201   //ETX  
00202   // Description:
00203   // Subroutine to update new cluster centers from the old centers.
00204   // Called from within Learn (and will be overridden by vtkPKMeansStatistics
00205   // to handle distributed datasets).
00206   virtual void UpdateClusterCenters( vtkTable* newClusterElements, 
00207                                      vtkTable* curClusterElements, 
00208                                      vtkIdTypeArray* numMembershipChanges,
00209                                      vtkIdTypeArray* numElementsInCluster, 
00210                                      vtkDoubleArray* error,
00211                                      vtkIdTypeArray* startRunID, 
00212                                      vtkIdTypeArray* endRunID, 
00213                                      vtkIntArray *computeRun );
00215 
00219   virtual vtkIdType GetTotalNumberOfObservations( vtkIdType numObservations );
00220 
00222 
00226   int InitializeDataAndClusterCenters(vtkTable* inParameters,
00227                                       vtkTable* inData,
00228                                       vtkTable*  dataElements,
00229                                       vtkIdTypeArray*  numberOfClusters,
00230                                       vtkTable*  curClusterElements,
00231                                       vtkTable*  newClusterElements,
00232                                       vtkIdTypeArray*  startRunID,
00233                                       vtkIdTypeArray*  endRunID);
00235 
00237 
00240   virtual void CreateInitialClusterCenters(vtkIdType numToAllocate, 
00241                                            vtkIdTypeArray* numberOfClusters, 
00242                                            vtkTable* inData, 
00243                                            vtkTable* curClusterElements, 
00244                                            vtkTable* newClusterElements);
00246 
00247 
00249 
00251   int DefaultNumberOfClusters;
00252   // Description:
00253   // This is the name of the column that specifies the number of clusters in each run. 
00254   // This is only used if the user has not specified initial clusters.
00255   char* KValuesArrayName;
00256   // Description:
00257   // This is the maximum number of iterations allowed if the new cluster centers have not yet converged. 
00258   int MaxNumIterations;
00259   // Description:
00260   // This is the percentage of data elements that swap cluster IDs 
00261   double Tolerance;
00262   // Description:
00263   // This is the Distance functor.  The default is Euclidean distance, however this can be overridden.
00264   vtkKMeansDistanceFunctor* DistanceFunctor;
00266   
00267 private:
00268   vtkKMeansStatistics( const vtkKMeansStatistics& ); // Not implemented
00269   void operator=( const vtkKMeansStatistics& );  // Not implemented
00270 };
00271 
00272 #endif