deal (version 1.2-4)

conditional: Calculate conditional distribution

Description

From master distribution the conditional distribution of a node given the parents is calculated.

Usage

conditional.cont(A,mu,nu,rho,phi) 
conditional(A,master,nw) 
conditional.disc(A,master) 
cond.node(node,nw,nw.prior=jointprior(nw))

Arguments

nw

an object of class network.

node

integer, giving the index of the current node.

nw.prior

a list describing parameter priors, generated by jointprior.

A

integer, giving the index of the current node.

master

a list describing the local master, generated by localmaster.

mu

numeric vector. Mu is a parameter in the local master.

nu

numeric. Nu is a parameter in the local master.

rho

numeric. Rho is a parameter in the local master.

phi

numeric matrix. Phi is a parameter in the local master.

Value

• The node is updated with condprior property, consisting of a list with components
• tau
  a numeric matrix, giving the unscaled precision matrix of regression parameters.
• mu
  a numeric vector, giving the mean of the regression parameters.
• phi
  a numeric, giving a parameter in the distribution of the scale parameter.
• rho
  a numeric, giving a parameter in the distribution of the scale parameter.
• alpha
  a numeric matrix with parameters for Dirichlet distribution.

Details

Used in learn, which calls cond.node. The local master is calculated and sent to conditional which calls either conditional.cont or conditional.disc depending on the type of the node. In each case the localmaster is used for calculating the conditional distribution of the current node given the remaining nodes in the local master.

References

Further information about deal can be found at: http://www.math.auc.dk/novo/deal.

See Also

learn, localmaster