| gmgm-package {gmgm} | R Documentation |
Gaussian mixture graphical model learning and inference
Description
This package provides a complete framework to deal with Gaussian mixture graphical models, which covers Bayesian networks and dynamic Bayesian networks (their temporal extension) whose local probability distributions are described by Gaussian mixture models. It includes a wide range of functions for:
creating or modifying the structure of Gaussian mixture models (
add_var,gmm,merge_comp,remove_var,rename_var,reorder,split_comp) or graphical models (add_arcs,add_nodes,gmbn,gmdbn,relevant,remove_arcs,remove_nodes,rename_nodes);describing or visualizing Gaussian mixture models (
conditional,ellipses,summary.gmm) or graphical models (network,structure,summary.gmbn,summary.gmdbn);computing densities, expectations, or sampling Gaussian mixture models (
density,expectation,sampling);computing scores of Gaussian mixture models (
AIC.gmm,BIC.gmm,logLik.gmm) or graphical models (AIC.gmbn,AIC.gmdbn,BIC.gmbn,BIC.gmdbn,logLik.gmbn,logLik.gmdbn);learning the structure and/or the parameters of Gaussian mixture models (
em,smem,stepwise) or graphical models (param_em,param_learn,struct_em,struct_learn);performing inference in Gaussian mixture graphical models (
aggregation,filtering,inference,particles,prediction,propagation,smoothing).
Descriptions of these functions are provided in this manual with related
references. Most of the algorithms are described in the PhD thesis of Roos
(2018, in french). To better handle this package, two real-world datasets are
provided (data_air, data_body) with examples of
Gaussian mixture models and graphical models (gmbn_body,
gmdbn_air, gmm_body).
References
Roos, J. (2018). Prévision a court terme des flux de voyageurs : une approche par les réseaux bayésiens. PhD thesis, University of Lyon.