Poisk2: Structure learning with Poisson models using the Poisson K2...

In drisso/learn2count: Structure Learning for Count Data

Structure learning with Poisson models using the Poisson K2 (PK2) algorithm

Description

This function finds the best fitting structure of a Poisson model given a matrix of counts and topological ordering, using a given criterion ("AIC", "BIC"). The PK2 algorithm is a modification of the K2 algorithm of Cooper and Herskovits (1992) able to deal with Poisson data. See Nguyen et al. (2022) for details.

Usage

Poisk2(X, order, criterion = "BIC", maxcard)

Arguments

X	the matrix of counts (n times p).
order	the topological ordering of variables (names of nodes).
criterion	the score function that measure the fitting of structures, could be "AIC" or "BIC".
maxcard	the uper bound of the cardinality of the parent sets.

Value

a list containing the estimated adjacency matrix of the graph and a graphNEL object of the same graph.

References

Cooper, G. F. and Herskovits, E. (1992). A Bayesian method for the induction of probabilistic networks from data. Machine learning, 9(4), 309–347.

Nguyen, Chiogna, Risso, Banzato (2022). Guided structure learning of DAGs for count data. arXiv:2206.09754.

drisso/learn2count documentation built on July 15, 2024, 11:13 p.m.