cibn-package: Causal Independence Bayesian Networks

In alessandromagrini/cibn: Causal Independence Bayesian Networks

Description

Elicitation, estimation and inference functionalities for Bayesian networks under the causal independence assumption.

Details

Package:	cibn
Type:	Package
Version:	0.0
Date:	2021-01-07
License:	GPL-2

Causal independence Bayesian networks (Magrini, 2021) are Bayesian networks with non-interacting parent variables (causal independence assumption). They allow three exaustive types of variables (graded, double-graded and multi-valued nominal variables) and admit the Causal Independence Decomposition (CID), which increases efficiency of elicitation, estimation and inference. Causal interactions can be added upon need. The main functions of the package are:

• new.cibn, to create a new network based on prior knowledge;

• update.cibn, to update an existing network based on possibly incomplete data (not still implemented but available soon);

• query.cibn, to perform exact inference in a network through an interface to the gRain package;

• sample.cibn, to draw a random sample from a network.

Author(s)

Alessandro Magrini <[e-mail]>

References

A. Magrini (2021). Efficient decomposition of Bayesian networks with non-graded variables. To be appeared on International Journal of Statistics and Probability, 10(2).

alessandromagrini/cibn documentation built on Feb. 7, 2022, 10:55 p.m.