BALL::QSAR::ALLModel Class Reference

#include <BALL/QSAR/allModel.h>

Inheritance diagram for BALL::QSAR::ALLModel:

Public Member Functions
Constructors and Destructors

	ALLModel (const QSARData &q, double kw=4)
	~ALLModel ()
Protected Attributes
Attributes

double	kw_
double	lambda_
Accessors

void	setKw (double kw)
virtual Vector< double >	predict (const vector< double > &substance, bool transform=1)
void	train ()
virtual bool	optimizeParameters (int d, int no_steps)
double	getKw ()
virtual void	setParameters (vector< double > &v)
virtual vector< double >	getParameters () const
void	saveToFile (string filename)
void	readFromFile (string filename)
virtual void	calculateWeights (BALL::Matrix< double > &dist, BALL::Vector< double > &w)
void	calculateXX (BALL::Vector< double > &w, BALL::Matrix< double > &res)
void	calculateXY (BALL::Vector< double > &w, BALL::Matrix< double > &res)
void	calculateEuclDistanceMatrix (BALL::Matrix< double > &m1, BALL::Matrix< double > &m2, BALL::Matrix< double > &output)

Detailed Description

class for automated lazy learning (ALL-QSAR)

Definition at line 40 of file allModel.h.

Constructor & Destructor Documentation

BALL::QSAR::ALLModel::ALLModel	(	const QSARData &	q,
		double	kw = `4`
	)

BALL::QSAR::ALLModel::~ALLModel ( )

Member Function Documentation

void BALL::QSAR::ALLModel::calculateEuclDistanceMatrix	(	BALL::Matrix< double > &	m1,
		BALL::Matrix< double > &	m2,
		BALL::Matrix< double > &	output
	)			`[protected]`

calculates pairwise euclidean distance between all substances of m1 and m2 and saves them to BALL::Matrix<double> output

virtual void BALL::QSAR::ALLModel::calculateWeights	(	BALL::Matrix< double > &	dist,
		BALL::Vector< double > &	w
	)			`[protected, virtual]`

void BALL::QSAR::ALLModel::calculateXX	(	BALL::Vector< double > &	w,
		BALL::Matrix< double > &	res
	)			`[protected]`

calculates weighted BALL::Matrix<double> X^T*X

void BALL::QSAR::ALLModel::calculateXY	(	BALL::Vector< double > &	w,
		BALL::Matrix< double > &	res
	)			`[protected]`

calculates weighted BALL::Matrix<double> X^T*Y

double BALL::QSAR::ALLModel::getKw ( )

returns the current kernel width

virtual vector<double> BALL::QSAR::ALLModel::getParameters ( ) const [virtual]

Reimplemented from BALL::QSAR::Model.

Reimplemented in BALL::QSAR::KNNModel.

virtual bool BALL::QSAR::ALLModel::optimizeParameters	(	int	d,
		int	no_steps
	)			`[virtual]`

Tries to find the optimal kernel width for the current data. Therefore some training data must have been read by the connected QSARData object before running this method.

Reimplemented from BALL::QSAR::Model.

Reimplemented in BALL::QSAR::KNNModel.

virtual Vector<double> BALL::QSAR::ALLModel::predict	(	const vector< double > &	substance,
		bool	transform = `1`
	)			`[virtual]`

Predicts the activities of a given substance

Parameters:

	substance	the substance which activity is to be predicted in form of a vecor containing the values for all descriptors (if neccessary, relevant descriptors will be selected automatically)
	transform	determines whether the values for each descriptor of the given substance should be transformed before prediction of activity. If (transform==1): each descriptor value is transformed according to the centering of the respective column of QSARData.descriptor_matrix used to train this model. If the substance to be predicted is part of the same input data (e.g. same SD-file) as the training data (as is the case during cross validation), transform should therefore be set to 0.