Manages Training and Test Data for building Statistical Models and provides functionality for Crossvalidation. More...

#include <DataManager.h>

Inheritance diagram for statismo::DataManager< T >:

Public Types
typedef Representer< T >	RepresenterType

typedef RepresenterType::DatasetPointerType	DatasetPointerType

typedef RepresenterType::DatasetConstPointerType	DatasetConstPointerType

typedef DataItem< T >	DataItemType

typedef DataItemWithSurrogates< T >	DataItemWithSurrogatesType

typedef std::list< const DataItemType * >	DataItemListType

typedef CrossValidationFold< T >	CrossValidationFoldType

typedef std::list < CrossValidationFoldType >	CrossValidationFoldListType

Public Member Functions
void	Delete ()

virtual	~DataManager ()

virtual void	AddDataset (DatasetConstPointerType dataset, const std::string &URI)

virtual void	Save (const std::string &filename) const

DataItemListType	GetData () const

unsigned	GetNumberOfSamples () const

CrossValidationFoldListType	GetCrossValidationFolds (unsigned nFolds, bool randomize=true) const

CrossValidationFoldListType	GetLeaveOneOutCrossValidationFolds () const

Static Public Member Functions
static DataManager< T > *	Create (const RepresenterType *representer)

static DataManager< T > *	Load (Representer< T > *representer, const std::string &filename)

Protected Member Functions
	DataManager (const RepresenterType *representer)

	DataManager (const DataManager< T > &orig)

DataManager &	operator= (const DataManager< T > &rhs)

Protected Attributes
RepresenterType *	m_representer

DataItemListType	m_DataItemList

Detailed Description

template<typename T>
class statismo::DataManager< T >

Manages Training and Test Data for building Statistical Models and provides functionality for Crossvalidation.

The DataManager class provides functionality for loading and managing data sets to be used in the statistical model. The datasets are loaded either by using DataManager::AddDataset or directly from a hdf5 File using the Load function. Per default all the datasets are marked as training data. It is, however, often useful to leave a few datasets out to validate the model. For this purpose, the DataManager class implements basic crossvalidation functionality.

Note that while Dataset are provided, the Representer class automatically converts them into Samples (Representer::DatasetToSample) For efficiency purposes, the data is internally stored as a large matrix, using the internal SampleVector representation (Representer::DatasetToSample). Furthermore, Statismo emphasizes on traceability, and ties information with the datasets, such as the original filename. This means that when accessing the data stored in the DataManager, one gets a DataItem structure

See Also: Representer; DataItem

Constructor & Destructor Documentation

template<typename T >

statismo::DataManager< T >::~DataManager ( )

virtual

Destructor

Member Function Documentation

template<typename T >

void statismo::DataManager< T >::AddDataset	(	DatasetConstPointerType	dataset,
		const std::string &	URI
	)

virtual

Add a dataset to the data manager.

Parameters

dataset	the dataset to be added
URI	A string containing the URI of the given dataset. This is only added as an info to the metadata.

While it is not strictly necessary, and sometimes not even possible, to specify a URI for the given dataset, it is strongly encouraged to add a description. The string will be added to the metadata and stored with the model. Having this information stored with the model may prove valuable at a later point in time.

template<typename T>

static DataManager<T>* statismo::DataManager< T >::Create ( const RepresenterType * representer )

inlinestatic

Factory method that creates a new instance of a DataManager class

template<typename T>

void statismo::DataManager< T >::Delete ( )

inline

Destroy the object. The same effect can be achieved by deleting the object in the usual way using the c++ delete keyword.

template<typename T >

DataManager< T >::CrossValidationFoldListType statismo::DataManager< T >::GetCrossValidationFolds	(	unsigned	nFolds,
		bool	randomize = `true`
	)		const

Assigns the data to one of n Folds to be used for cross validation. This method has to be called before cross validation can be started.

Parameters

nFolds	The number of folds used in the crossvalidation
randomize	If true, the data will be randomly assigned to the nfolds, otherwise the order with which it was added is preserved

template<typename T >

DataManager< T >::DataItemListType statismo::DataManager< T >::GetData ( ) const

return a list with all the sample data objects managed by the data manager

See Also: DataItem

template<typename T >

DataManager< T >::CrossValidationFoldListType statismo::DataManager< T >::GetLeaveOneOutCrossValidationFolds ( ) const

Generates Leave-one-out cross validation folds

template<typename T>

unsigned statismo::DataManager< T >::GetNumberOfSamples ( ) const

inline

returns the number of samples managed by the datamanager

template<typename T>

DataManager< T > * statismo::DataManager< T >::Load	(	Representer< T > *	representer,
		const std::string &	filename
	)

static

Create a new dataManager, with the data stored in the given hdf5 file

template<typename T >

void statismo::DataManager< T >::Save ( const std::string & filename ) const

virtual

Saves the data matrix and all URIs into an HDF5 file.

Parameters

filename

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

Public Types

Public Member Functions

Static Public Member Functions

Protected Member Functions

Protected Attributes

Detailed Description

template<typename T> class statismo::DataManager< T >

Constructor & Destructor Documentation

Member Function Documentation

template<typename T>
class statismo::DataManager< T >