BALL::QSAR::FitModel Class Reference

#include <BALL/QSAR/fitModel.h>

Inheritance diagram for BALL::QSAR::FitModel:
Inheritance graph
[legend]

List of all members.

Public Member Functions

Constructors and Destructors

 FitModel (const QSARData &q)
 FitModel (const QSARData &q, vector< String > &eq, vector< double > &guess)
 FitModel (const QSARData &q, vector< String > &eq, vector< vector< String > > &deq, vector< double > &guess)
Accessors

void setEquations (vector< String > &eq, vector< vector< String > > &deq)
void setEquations (vector< String > &eq)
void train ()
Vector< doublepredict (const vector< double > &substance, bool transform=1)
void setInitialGuess (vector< double > &d)

Private Attributes

Attributes

vector< StringallEquations_
vector< vector< String > > allDiffEquations_
vector< doubleinitial_guess_

Detailed Description

Definition at line 44 of file fitModel.h.


Constructor & Destructor Documentation

BALL::QSAR::FitModel::FitModel ( const QSARData q  ) 
BALL::QSAR::FitModel::FitModel ( const QSARData q,
vector< String > &  eq,
vector< double > &  guess 
)
BALL::QSAR::FitModel::FitModel ( const QSARData q,
vector< String > &  eq,
vector< vector< String > > &  deq,
vector< double > &  guess 
)

Member Function Documentation

Vector<double> BALL::QSAR::FitModel::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.
Returns:
a RowVector containing one value for each predicted activity

Implements BALL::QSAR::Model.

void BALL::QSAR::FitModel::setEquations ( vector< String > &  eq  ) 

sets all equations but no derivatives. Since no derivatives are given, numerical derivation will be used during training.

void BALL::QSAR::FitModel::setEquations ( vector< String > &  eq,
vector< vector< String > > &  deq 
)

sets all equations and their derivatives

Parameters:
eq vector containing one equation for each column of Y
deq derivatives of each equation in one vector
void BALL::QSAR::FitModel::setInitialGuess ( vector< double > &  d  ) 
void BALL::QSAR::FitModel::train (  )  [virtual]

starts training the FitModel using Levenberg-Marquardt

Implements BALL::QSAR::Model.


Member Data Documentation

vector<vector<String> > BALL::QSAR::FitModel::allDiffEquations_ [private]

saves all derivatives of one equation in one vector

Definition at line 88 of file fitModel.h.

saves one equations for the prediction of each activitiy, i.e. one String for each column of Y

Definition at line 85 of file fitModel.h.

saves the approximated coefficients for each equation, i.e. one String for each column of Y

Definition at line 93 of file fitModel.h.

 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Properties Friends Defines
Generated by  doxygen 1.6.3