trilearn_hyper-dir

Hyper-Dirichlet

Package	trilearn
Version	1.25
Language	Python
Docs
Paper	Dawid and Lauritzen[1]
Graph type	DG
Docker	bpimages/trilearn:2.0.5
Module	trilearn_hyper-dir

Description

An object of this module defines a categorical decomposable model by sampling the parameters from the hyper Dirichlet distribution, with specified equivalent sample size (pseudo_obs) and number of levels per variable (n_levels).

Note

This module currently only supports graphs with up to 15 nodes.

Example

[
  {
    "id": "disc-loglin",
    "n_levels": 2,
    "pseudo_obs": 1.0
  }
]