CvHanler is an utility function for building the submatrix/subvectors needed when using k-folds cross-validation. More...

#include <STK_CvHandler.h>

Inheritance diagram for STK::CvHandler:

Public Member Functions
	CvHandler (Range const &rangeData, int nbFolds)
	Default constructor.

virtual	~CvHandler ()
	destructor

int	nbFolds () const

Range const &	rangeData () const

CVectorXi const &	partitions () const

CVectorXi const &	sizePartitions () const

virtual bool	run ()
	run the computations.

void	setData (Range const &rangeData, int nbFolds)

template<class Data >
bool	getKFold (int k, Data const &x, Data &xFold, Data &xTest)
	get the data set when setting out fold k and test data set

template<class xData , class yData >
bool	getKFold (int k, xData const &x, xData &xFold, xData &xTest, yData const &y, yData &yFold, yData &yTest)
	get the data set when setting out fold k and test data set

Public Member Functions inherited from STK::IRunnerBase
String const &	error () const
	get the last error message.

Protected Member Functions
void	partition ()
	create a random partition in k folds

Protected Member Functions inherited from STK::IRunnerBase
	IRunnerBase ()
	default constructor

	IRunnerBase (IRunnerBase const &runner)
	copy constructor

virtual	~IRunnerBase ()
	destructor

virtual void	update ()
	update the runner.

Private Attributes
Range	rangeData_
	Range of the data set (number of rows)

int	nbFolds_
	Number of folds.

CVectorXi	partitions_
	repartition of the sample into k-folds

CVectorXi	sizePartitions_
	size of each fold

Additional Inherited Members
Protected Attributes inherited from STK::IRunnerBase
String	msg_error_
	String with the last error message.

bool	hasRun_
	`true` if run has been used, `false` otherwise

Detailed Description

CvHanler is an utility function for building the submatrix/subvectors needed when using k-folds cross-validation.

Definition at line 52 of file STK_CvHandler.h.

Constructor & Destructor Documentation

◆ CvHandler()

STK::CvHandler::CvHandler	(	Range const &	rangeData,
		int	nbFolds
	)

inline

Default constructor.

Parameters

rangeData,nbFolds range of the data and number of folds

Definition at line 109 of file STK_CvHandler.h.

                           : IRunnerBase()
                           , rangeData_(rangeData), nbFolds_(nbFolds)
                           , partitions_(), sizePartitions_()
{
  // check nbFolds parameter
  if (nbFolds_<1)
  { STKRUNTIME_ERROR_1ARG(CvHandler::CvHandler,nbFolds,nbFolds<1);}
  if (nbFolds_>rangeData_.size())
  { STKRUNTIME_ERROR_1ARG(CvHandler::CvHandler,nbFolds,nbFolds>rangeData_.size());}
}

References CvHandler(), nbFolds(), nbFolds_, rangeData_, STK::TRange< UnknownSize >::size(), and STKRUNTIME_ERROR_1ARG.

Referenced by CvHandler().

◆ ~CvHandler()

virtual STK::CvHandler::~CvHandler ( )

inlinevirtual

destructor

Definition at line 60 of file STK_CvHandler.h.

60{}

Member Function Documentation

◆ getKFold() [1/2]

template<class Data >

bool STK::CvHandler::getKFold	(	int	k,
		Data const &	x,
		Data &	xFold,
		Data &	xTest
	)

get the data set when setting out fold k and test data set

Definition at line 122 of file STK_CvHandler.h.

{
  // check if partitions are determined
  if (!hasRun_)
  { msg_error_ = STKERROR_NO_ARG(CvHandler::getKFold,CvHandler has to run);
    return false;
  }
  // check dimensions
  if (x.rows() != rangeData_)
  { msg_error_ = STKERROR_1ARG(CvHandler::getKFold,k,x.rows()!=rangeData_);
    return false;
  }
  if (sizePartitions_.begin() > k)
  { msg_error_ = STKERROR_1ARG(CvHandler::getKFold,k,k<sizePartitions_.begin());
    return false;
  }
  if (sizePartitions_.end() <= k)
  { msg_error_ = STKERROR_1ARG(CvHandler::getKFold,k,k<sizePartitions_.end()<=k);
    return false;
  }
  // prepare containers
  Range xFoldRows = x.rows();
  xFoldRows.decLast(sizePartitions_[k]);
  xFold.resize(xFoldRows, x.cols());
  xTest.resize(sizePartitions_[k], x.cols());
  // copy data
  int iFoldRow = xFold.beginRows(), iTestRow = xTest.beginRows();
  for (int i = partitions_.begin(); i < partitions_.end(); ++i)
  {
    if (partitions_[i] == k)
    {
      xTest.row(iTestRow) = x.row(i);
      ++iTestRow;
    }
    else
    {
      xFold.row(iFoldRow) = x.row(i);
      ++iFoldRow;
    }
  }
  return true;
}

References getKFold(), STK::IRunnerBase::hasRun_, STK::IRunnerBase::msg_error_, partitions_, rangeData_, run(), sizePartitions_, STKERROR_1ARG, and STKERROR_NO_ARG.

Referenced by getKFold(), and getKFold().

◆ getKFold() [2/2]

template<class xData , class yData >

bool STK::CvHandler::getKFold	(	int	k,
		xData const &	x,
		xData &	xFold,
		xData &	xTest,
		yData const &	y,
		yData &	yFold,
		yData &	yTest
	)

get the data set when setting out fold k and test data set

Definition at line 166 of file STK_CvHandler.h.

{
  // check if partitions are determined
  if (!hasRun_)
  { msg_error_ = STKERROR_NO_ARG(CvHandler::getKFold,CvHandler has to run);
    return false;
  }
  // check dimensions
  if (x.rows() != rangeData_)
  { msg_error_ = STKERROR_1ARG(CvHandler::getKFold,k,x.rows()!=rangeData_);
    return false;
  }
  if (y.rows() != rangeData_)
  { msg_error_ = STKERROR_1ARG(CvHandler::getKFold,k,y.rows()!=rangeData_);
    return false;
  }
  if (sizePartitions_.begin() > k)
  { msg_error_ = STKERROR_1ARG(CvHandler::getKFold,k,k<sizePartitions_.begin());
    return false;
  }
  if (sizePartitions_.end() <= k)
  { msg_error_ = STKERROR_1ARG(CvHandler::getKFold,k,k<sizePartitions_.end()<=k);
    return false;
  }
  // prepare constainers
  Range xFoldRows = x.rows();
  xFoldRows.decLast(sizePartitions_[k]);
  xFold.resize(xFoldRows, x.cols());
  xTest.resize(sizePartitions_[k], x.cols());
  yFold.resize(xFoldRows, y.cols());
  yTest.resize(sizePartitions_[k], y.cols());
  // copy data
  int iFoldRow = xFold.beginRows(), iTestRow = xTest.beginRows();
  for (int i = partitions_.begin(); i < partitions_.end(); ++i)
  {
    if (partitions_[i] == k)
    {
      xTest.row(iTestRow) = x.row(i);
      yTest.row(iTestRow) = y.row(i);
      ++iTestRow;
    }
    else
    {
      xFold.row(iFoldRow) = x.row(i);
      yFold.row(iFoldRow) = y.row(i);
      ++iFoldRow;
    }
  }
  return true;
}

References getKFold(), STK::IRunnerBase::hasRun_, STK::IRunnerBase::msg_error_, partitions_, rangeData_, run(), sizePartitions_, STKERROR_1ARG, and STKERROR_NO_ARG.

◆ nbFolds()

int STK::CvHandler::nbFolds ( ) const

inline

Returns: the number of folds

Definition at line 63 of file STK_CvHandler.h.

63{ return nbFolds_;}

References nbFolds_.

Referenced by CvHandler(), and setData().

◆ partition()

void STK::CvHandler::partition ( )

inlineprotected

create a random partition in k folds

Definition at line 219 of file STK_CvHandler.h.

{
  partitions_.resize(rangeData_);
  sizePartitions_.resize(Range(0, nbFolds_)) = 0;
  //fill the container with the index of folds
  for(int i = partitions_.begin() ; i< partitions_.end() ;i++)
  {
    partitions_[i] = i%nbFolds_;
    sizePartitions_[i%nbFolds_]++;
  }
  //make a random rearrangement
  int begin = partitions_.begin();
  for (int i=partitions_.end()-2; i>begin; --i)
  { std::swap(partitions_[i], partitions_[Law::UniformDiscrete::rand(begin, i+1)]);}
}

References nbFolds_, partitions_, STK::Law::UniformDiscrete::rand(), rangeData_, STK::ICArray< Derived >::resize(), and sizePartitions_.

Referenced by run().

◆ partitions()

CVectorXi const & STK::CvHandler::partitions ( ) const

inline

Returns: the partitions

Definition at line 67 of file STK_CvHandler.h.

67{ return partitions_;}

References partitions_.

◆ rangeData()

Range const & STK::CvHandler::rangeData ( ) const

inline

Returns: the range of the data

Definition at line 65 of file STK_CvHandler.h.

65{ return rangeData_;}

References rangeData_.

Referenced by setData().

◆ run()

virtual bool STK::CvHandler::run ( )

inlinevirtual

run the computations.

Returns: true if no error occur during the running process, false otherwise

Implements STK::IRunnerBase.

Definition at line 71 of file STK_CvHandler.h.

72 { partition(); hasRun_ = true; return true;}

STK::CvHandler::partition

void partition()

create a random partition in k folds

Definition STK_CvHandler.h:219

References STK::IRunnerBase::hasRun_, and partition().

Referenced by getKFold(), and getKFold().

◆ setData()

void STK::CvHandler::setData	(	Range const &	rangeData,
		int	nbFolds
	)

inline

Definition at line 74 of file STK_CvHandler.h.

    {
      rangeData_ = (rangeData);
      nbFolds_ = nbFolds;
      partitions_.clear();
      sizePartitions_.clear();
      hasRun_ = false;
    }

References STK::ICArray< Derived >::clear(), STK::IRunnerBase::hasRun_, nbFolds(), nbFolds_, partitions_, rangeData(), rangeData_, and sizePartitions_.

◆ sizePartitions()

CVectorXi const & STK::CvHandler::sizePartitions ( ) const

inline

Returns: the size of the partitions

Definition at line 69 of file STK_CvHandler.h.

69{ return sizePartitions_;}

References sizePartitions_.

Member Data Documentation

◆ nbFolds_

int STK::CvHandler::nbFolds_

private

Number of folds.

Definition at line 98 of file STK_CvHandler.h.

Referenced by CvHandler(), nbFolds(), partition(), and setData().

◆ partitions_

CVectorXi STK::CvHandler::partitions_

private

repartition of the sample into k-folds

Definition at line 100 of file STK_CvHandler.h.

Referenced by getKFold(), getKFold(), partition(), partitions(), and setData().

◆ rangeData_

Range STK::CvHandler::rangeData_

private

Range of the data set (number of rows)

Definition at line 96 of file STK_CvHandler.h.

Referenced by CvHandler(), getKFold(), getKFold(), partition(), rangeData(), and setData().

◆ sizePartitions_

CVectorXi STK::CvHandler::sizePartitions_

private

size of each fold

Definition at line 102 of file STK_CvHandler.h.

Referenced by getKFold(), getKFold(), partition(), setData(), and sizePartitions().

The documentation for this class was generated from the following file:

STK_CvHandler.h

Public Member Functions

Protected Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ CvHandler()

◆ ~CvHandler()

Member Function Documentation

◆ getKFold() [1/2]

◆ getKFold() [2/2]

◆ nbFolds()

◆ partition()

◆ partitions()

◆ rangeData()

◆ run()

◆ setData()

◆ sizePartitions()

Member Data Documentation

◆ nbFolds_

◆ partitions_

◆ rangeData_

◆ sizePartitions_