cp.death: Make changepoint death move.

In EDISON: Network Reconstruction and Changepoint Detection

Description

This function makes a changepoint death move, possibly removing a changepoint.

Usage

cp.death(Eall, Sall, Ball, Sig2all, X, Y, D, GLOBvar, HYPERvar, target)

Arguments

Eall	Changepoints: List of target nodes, where each element contains a vector of changepoints.
Sall	Network structure: List of target nodes, where each element is a NumSegs by NumNodes matrix giving the parents for the target node in each segment. A binary matrix.
Ball	Network parameters: Similar to network structure, but with regression parameters included.
Sig2all	Sigma squared parameters.
X	Response data.
Y	Target data.
D	Hyperparameter.
GLOBvar	Global variables of the MCMC simulation.
HYPERvar	Hyperparameter variables.
target	Which target node the move is being proposed for.

Value

A list with elements:

E	New changepoint vector for target node.
Sall	Updated network structure.
Ball	Updated network structure with regression parameters.
Sig2all	Updated sigma squared.
prior.params	Updated vector of structure prior hyperparameters.
accept	Whether the move was accepted or not.
move	What type of move was made. In this case move=2 for a changepoint death move.
alpha	The acceptance ratio of the move.
estar	The location of the removed changepoint.
k	Hyperparameter.

Author(s)

Sophie Lebre

Frank Dondelinger

References

For more information about the different changepoint moves, see:

Dondelinger et al. (2012), "Non-homogeneous dynamic Bayesian networks with Bayesian regularization for inferring gene regulatory networks with gradually time-varying structure", Machine Learning.

See Also

cp.birth, cp.shift

EDISON documentation built on May 2, 2019, 2:39 a.m.