UnlabelledData.h

Go to the documentation of this file.

#ifndef GRT_UNLABLELLED_DATA_HEADER
#define GRT_UNLABLELLED_DATA_HEADER

#include "../Util/GRTCommon.h"

GRT_BEGIN_NAMESPACE

class GRT_API UnlabelledData{
public:
    
    UnlabelledData(const UINT numDimensions = 0,const std::string datasetName = "NOT_SET",const std::string infoText = "");

 UnlabelledData(const UnlabelledData &rhs);

 ~UnlabelledData();

 UnlabelledData& operator= (const UnlabelledData &rhs);

 inline VectorFloat& operator[] (const UINT &i){
  return data[i];
 }
    
 const inline VectorFloat& operator[] (const UINT &i) const{
  return data[i];
 }

 void clear();
    
    bool setNumDimensions(const UINT numDimensions);
    
    bool setDatasetName(const std::string datasetName);
    
    bool setInfoText(const std::string infoText);

 bool addSample(const VectorFloat &sample);
    
 bool removeLastSample();
    
    bool reserve(const UINT N);
    
    bool setExternalRanges(const Vector< MinMax > &externalRanges, const bool useExternalRanges = false);
    
    bool enableExternalRangeScaling(const bool useExternalRanges);

    bool scale(const Float minTarget,const Float maxTarget);
    
 bool scale(const Vector<MinMax> &ranges,const Float minTarget,const Float maxTarget);
    
    bool save(const std::string &filename) const;
    
    bool load(const std::string &filename);
 
 bool saveDatasetToFile(const std::string &filename) const;
 
 bool loadDatasetFromFile(const std::string &filename);
    
 bool saveDatasetToCSVFile(const std::string &filename) const;
 
 bool loadDatasetFromCSVFile(const std::string &filename);

 GRT_DEPRECATED_MSG( "partition(...) is deprecated, use split(...) instead", UnlabelledData partition(const UINT partitionPercentage) );

    UnlabelledData split(const UINT partitionPercentage);
    
    bool merge(const UnlabelledData &unlabelledData);
    
    bool spiltDataIntoKFolds(const UINT K);
    
    UnlabelledData getTrainingFoldData(const UINT foldIndex) const;
    
    UnlabelledData getTestFoldData(const UINT foldIndex) const;
    
    std::string getDatasetName() const{ return datasetName; }
    
    std::string getInfoText() const{ return infoText; }

    std::string getStatsAsString() const;
    
 UINT inline getNumDimensions() const{ return numDimensions; }
 
 UINT inline getNumSamples() const{ return totalNumSamples; }
 
 Vector<MinMax> getRanges() const;
    
 Vector< VectorFloat > getData() const;
    
 MatrixDouble getDataAsMatrixDouble() const;

    MatrixFloat getDataAsMatrixFloat() const;

private:
    std::string datasetName;                                
    std::string infoText;                                   
 UINT numDimensions;          
 UINT totalNumSamples;                                   
    UINT kFoldValue;                                        
    bool crossValidationSetup;                              
    bool useExternalRanges;                                 
    Vector< MinMax > externalRanges;                        
 
 Vector< VectorFloat > data;                            
    Vector< Vector< UINT > >    crossValidationIndexs;      
    
    DebugLog debugLog;                                      
    ErrorLog errorLog;                                      
    WarningLog warningLog;                                  
        
};

GRT_END_NAMESPACE

#endif //GRT_UNLABLELLED_CLASSIFICATION_DATA_HEADER