BALL::QSAR::Validation Class Reference

#include <BALL/QSAR/validation.h>

Inheritance diagram for BALL::QSAR::Validation:

Public Member Functions
Constructors and Destructors
	Validation (Model *m)
virtual	~Validation ()

Protected Attributes
Attributes
vector< vector< double > >	test_substances_
Eigen::MatrixXd	test_Y_
vector< string >	substances_names_
Model *	model_
int	validation_statistic_
Eigen::MatrixXd	yRand_results_

Accessors
virtual void	crossValidation (int k, bool restore=1)=0
virtual double	getCVRes ()=0
virtual double	getFitRes ()=0
virtual void	setCVRes (double d)=0
const Eigen::MatrixXd &	getYRandResults () const
virtual void	testInputData (bool transform=0)=0
virtual void	bootstrap (int k, bool restore=1)=0
virtual const Eigen::MatrixXd &	yRandomizationTest (int runs, int k)=0
virtual void	selectStat (int s)=0
int	getStat () const
virtual void	saveToFile (string filename) const =0
virtual void	readFromFile (string filename)=0
void	setTrainingLine (int train_line, int current_line)
void	setTestLine (int test_line, int current_line, bool back_transform=0)
void	yRand ()

Detailed Description

interface class for validation of QSAR-models

Definition at line 23 of file validation.h.

Constructor & Destructor Documentation

BALL::QSAR::Validation::Validation

(

Model *

m

)

virtual BALL::QSAR::Validation::~Validation ( )

virtual

Member Function Documentation

virtual void BALL::QSAR::Validation::bootstrap ( int  k, bool  restore = 1  )

pure virtual

Implemented in BALL::QSAR::ClassificationValidation, and BALL::QSAR::RegressionValidation.

virtual void BALL::QSAR::Validation::crossValidation ( int  k, bool  restore = 1  )

pure virtual

Starts cross-validation with k steps.
Data is taken from QSARData.descriptor_matrix and is in each step divided into training- and test-data.
(Data having already been copied into Model.descriptor_matrix will be deleted)

Implemented in BALL::QSAR::RegressionValidation, and BALL::QSAR::ClassificationValidation.

virtual double BALL::QSAR::Validation::getCVRes ( )

pure virtual

fetches the result of cross-validation

Implemented in BALL::QSAR::RegressionValidation, and BALL::QSAR::ClassificationValidation.

virtual double BALL::QSAR::Validation::getFitRes ( )

pure virtual

fetches the quality of fit to the input data, as calculated by testInputData()

Implemented in BALL::QSAR::RegressionValidation, and BALL::QSAR::ClassificationValidation.

int BALL::QSAR::Validation::getStat

(

)

const

return the ID of selected validation statistic

const Eigen::MatrixXd& BALL::QSAR::Validation::getYRandResults

(

)

const

return the result of the previous response permutation test .
The return matrix is empty if no such test has been run yet

virtual void BALL::QSAR::Validation::readFromFile ( string  filename )

pure virtual

restore validation-results from a file

Implemented in BALL::QSAR::RegressionValidation, and BALL::QSAR::ClassificationValidation.

virtual void BALL::QSAR::Validation::saveToFile ( string  filename ) const

pure virtual

save the result of the applied validation methods to a file

Implemented in BALL::QSAR::RegressionValidation, and BALL::QSAR::ClassificationValidation.

virtual void BALL::QSAR::Validation::selectStat ( int  s )

pure virtual

select the desired statistic to be used for validating the models

Parameters

s	if (s==1) R^2 and Q^2 are used if(s==2) F_regr and F_cv are used.

Implemented in BALL::QSAR::RegressionValidation, and BALL::QSAR::ClassificationValidation.

virtual void BALL::QSAR::Validation::setCVRes ( double  d )

pure virtual

set the result of cross-validation to the given value

Implemented in BALL::QSAR::RegressionValidation, and BALL::QSAR::ClassificationValidation.

void BALL::QSAR::Validation::setTestLine ( int  test_line, int  current_line, bool  back_transform = 0  )

protected

copies one substance (=one line from class QSARData) and its activity into the test data set

Parameters

back_transform

descriptor values are transformed back to in order to un-do centering

void BALL::QSAR::Validation::setTrainingLine ( int  train_line, int  current_line  )

protected

copies one substance (=one line from class QSARData) and its activity into the training data set

virtual void BALL::QSAR::Validation::testInputData ( bool  transform = 0 )

pure virtual

Fetches input data from QSARData and tests the current (unchanged) model with all these new substances (without cross-validation!).

Parameters

transform

if transform==1, the test data is transformed in the same way that the training data was transformed before predicting activities. If training and test substances are taken from the same input file, set transform to 0

Implemented in BALL::QSAR::RegressionValidation, and BALL::QSAR::ClassificationValidation.

void BALL::QSAR::Validation::yRand ( )

protected

randomizes all columns of Model.Y_

virtual const Eigen::MatrixXd& BALL::QSAR::Validation::yRandomizationTest ( int  runs, int  k  )

pure virtual

Implemented in BALL::QSAR::RegressionValidation, and BALL::QSAR::ClassificationValidation.

Member Data Documentation

Model* BALL::QSAR::Validation::model_

protected

pointer to the model, which the object of this class should test

Definition at line 114 of file validation.h.

vector<string> BALL::QSAR::Validation::substances_names_

protected

contains the names of all current test-substances

Definition at line 110 of file validation.h.

vector<vector<double> > BALL::QSAR::Validation::test_substances_

protected

contains all substances whose activities are to be predicted by the next call of testAllSubstances()

Definition at line 104 of file validation.h.

Eigen::MatrixXd BALL::QSAR::Validation::test_Y_

protected

contains the experimentally determined results (activities) for all current test-substances in one column for each activity

Definition at line 107 of file validation.h.

int BALL::QSAR::Validation::validation_statistic_

protected

the selected validation statistic

Definition at line 117 of file validation.h.

Eigen::MatrixXd BALL::QSAR::Validation::yRand_results_

protected

Definition at line 119 of file validation.h.