#include <BALL/QSAR/bayesModel.h>
Public Member Functions | |
BayesModel (const QSARData &q) | |
virtual bool | isTrained ()=0 |
virtual vector< double > | calculateProbabilities (int activitiy_index, int feature_index, double feature_value)=0 |
virtual int | getNoResponseVariables ()=0 |
Protected Attributes | |
double | min_prob_diff_ |
double | undef_act_class_id_ |
interface class for all Bayes classification models
Definition at line 38 of file bayesModel.h.
BALL::QSAR::BayesModel::BayesModel | ( | const QSARData & | q | ) |
virtual vector<double> BALL::QSAR::BayesModel::calculateProbabilities | ( | int | activitiy_index, | |
int | feature_index, | |||
double | feature_value | |||
) | [pure virtual] |
calculate the probability for the specified feature to assume the given value for each class
Implemented in BALL::QSAR::NBModel, and BALL::QSAR::SNBModel.
virtual int BALL::QSAR::BayesModel::getNoResponseVariables | ( | ) | [pure virtual] |
returns the number of response variables for which this model has been trained
Implemented in BALL::QSAR::NBModel, and BALL::QSAR::SNBModel.
virtual bool BALL::QSAR::BayesModel::isTrained | ( | ) | [pure virtual] |
Implemented in BALL::QSAR::NBModel, and BALL::QSAR::SNBModel.
double BALL::QSAR::BayesModel::min_prob_diff_ [protected] |
the minimal desired difference between the predicted class and the second best class. If the difference is smaller than this value, undef_act_class_id_ should be used instead.
Definition at line 56 of file bayesModel.h.
double BALL::QSAR::BayesModel::undef_act_class_id_ [protected] |
the class_id that is to be assigned, if no class could be savely predicted, if difference between best and second best potential class is smaller than min_prob_diff_
Definition at line 59 of file bayesModel.h.