 |
aGrUM 2.3.2
a C++ library for (probabilistic) graphical models
|
|
aGrUM 2.3.2
a C++ library for (probabilistic) graphical models
|
the internal prior for the BDeu score (N' / (r_i * q_i) More...
#include <agrum/base/database/bdeuPrior.h>
Public Member Functions | |
Constructors / Destructors | |
| BDeuPrior (const DatabaseTable &database, const Bijection< NodeId, std::size_t > &nodeId2columns=Bijection< NodeId, std::size_t >()) | |
| default constructor | |
| BDeuPrior (const BDeuPrior &from) | |
| copy constructor | |
| BDeuPrior (BDeuPrior &&from) noexcept | |
| move constructor | |
| BDeuPrior * | clone () const override |
| virtual copy constructor | |
| virtual | ~BDeuPrior () |
| destructor | |
Operators | |
| BDeuPrior & | operator= (const BDeuPrior &from) |
| copy operator | |
| BDeuPrior & | operator= (BDeuPrior &&from) noexcept |
| move operator | |
Accessors / Modifiers | |
| void | setWeight (double weight) final |
| sets the effective sample size N' (alias of setEffectiveSampleSize ()) | |
| void | setEffectiveSampleSize (double weight) |
| sets the effective sample size N' | |
| PriorType | getType () const final |
| returns the type of the prior | |
| bool | isInformative () const final |
| indicates whether the prior is tensorly informative | |
| void | addJointPseudoCount (const IdCondSet &idset, std::vector< double > &counts) final |
| adds the prior to a counting vector corresponding to the idset | |
| void | addConditioningPseudoCount (const IdCondSet &idset, std::vector< double > &counts) final |
| adds the prior to a counting vector defined over the right hand side of the idset | |
Accessors / Modifiers | |
| double | weight () const |
| returns the weight assigned to the prior | |
Protected Attributes | |
| double | weight_ {1.0} |
| the weight of the prior | |
| const DatabaseTable * | database_ |
| a reference to the database in order to have access to its variables | |
| Bijection< NodeId, std::size_t > | nodeId2columns_ |
| a mapping from the NodeIds of the variables to the indices of the columns in the database | |
the internal prior for the BDeu score (N' / (r_i * q_i)
BDeu is a BD score with a N'/(r_i * q_i) prior, where N' is an effective sample size and r_i is the domain size of the target variable and q_i is the domain size of the Cartesian product of its parents.
It is important to note that, to be meaningful a structure + parameter learning requires that the same priors are taken into account during structure learning and parameter learning.
Definition at line 71 of file bdeuPrior.h.
|
explicit |
default constructor
| database | the database from which learning is performed. This is useful to get access to the random variables |
| nodeId2Columns | a mapping from the ids of the nodes in the graphical model to the corresponding column in the DatabaseTable. This enables estimating from a database in which variable A corresponds to the 2nd column the parameters of a BN in which variable A has a NodeId of 5. An empty nodeId2Columns bijection means that the mapping is an identity, i.e., the value of a NodeId is equal to the index of the column in the DatabaseTable. |
Referenced by BDeuPrior(), BDeuPrior(), clone(), operator=(), and operator=().
| gum::learning::BDeuPrior::BDeuPrior | ( | const BDeuPrior & | from | ) |
|
noexcept |
|
virtual |
destructor
|
finalvirtual |
adds the prior to a counting vector defined over the right hand side of the idset
Implements gum::learning::Prior.
References addConditioningPseudoCount().
Referenced by addConditioningPseudoCount().
|
finalvirtual |
adds the prior to a counting vector corresponding to the idset
adds the prior to an already created counting vector defined over the union of the variables on both the left and right hand side of the conditioning bar of the idset.
Implements gum::learning::Prior.
References addJointPseudoCount().
Referenced by addJointPseudoCount().
|
overridevirtual |
virtual copy constructor
Implements gum::learning::Prior.
References BDeuPrior().
|
finalvirtual |
returns the type of the prior
Implements gum::learning::Prior.
|
finalvirtual |
indicates whether the prior is tensorly informative
Basically, only the NoPrior is uninformative. However, it may happen that, under some circumstances, an prior, which is usually not equal to the NoPrior, becomes equal to it (e.g., when the weight is equal to zero). In this case, if the prior can detect this case, it shall inform the classes that use it that it is temporarily uninformative. These classes will then be able to speed-up their code by avoiding to take into account the prior in their computations.
Implements gum::learning::Prior.
References isInformative().
Referenced by isInformative().
| void gum::learning::BDeuPrior::setEffectiveSampleSize | ( | double | weight | ) |
sets the effective sample size N'
References gum::learning::Prior::weight().
|
finalvirtual |
sets the effective sample size N' (alias of setEffectiveSampleSize ())
Reimplemented from gum::learning::Prior.
References gum::learning::Prior::weight().
|
inherited |
returns the weight assigned to the prior
Referenced by gum::learning::BDeuPrior::setEffectiveSampleSize(), gum::learning::BDeuPrior::setWeight(), gum::learning::DirichletPriorFromBN< GUM_SCALAR >::setWeight(), gum::learning::DirichletPriorFromDatabase::setWeight(), gum::learning::K2Prior::setWeight(), gum::learning::NoPrior::setWeight(), and setWeight().
|
protectedinherited |
|
protectedinherited |