Bayesian Network

From COSSAN Wiki
Jump to: navigation, search

This is a probabilistic model presented in a graphical way by using Directed Acyclic Diagrams to study the probabilistic relationships among the nodes that make up the network. Each node represent a random variable.

The current version only accepts discrete nodes.

Tutorial

echodemo TutorialBayesianNetwork


Nodes definition

Provide the property values required to define the node as follows:

Property Description
Sdescription Description of the object
Sname Name of the node
CSparents Cell array of parent nodes
CSchildren Cell array of children nodes
CPD Conditional probability table
Chyperparameters Cell of intervals (hyperparameters) SAME SIZE OF CPD
Cmapping Input mapping for the interval random variables
Nsize Size of the node
Lroot =false Flag for root nodes
Lboolean= false Flag for Boolean nodes (only 2 possible outcomes: true and false)
Stype Discrete/Continuous/Bounded?
evidence Value of the evidence introduced
Vvalues Values assumed by the parameter (only for discrete nodes spouses of non-discrete nodes)
Vbounds Vector of bounds of discrete nodes
censoring=[] value of censoring: in case of construction of UserDefRV the realizations equal or below the censoring value are excluded from the distribution (conditional distribution)
MimportantDirection Important direction for the extreme case computation: N.B. each row is associated to a state of the node
Lpbounds=false Flag


Name the node with only one word, e.g., 'Rain'. If it is necessary to use more than one word, separate each word with the underscore sign "_", e.g., 'Wet_Grass'. Forgetting using the underscore sign to separate words might lead to errors when executing the code.