simulateDag: Simulate a DAG with with arbitrary arcs density

In abn: Modelling Multivariate Data with Additive Bayesian Networks

Simulate a DAG with with arbitrary arcs density

Description

Provided with node names, returns an abnDAG. Arc density refers to the chance of a node being connected to the node before it.

Usage

simulateDag(node.name, data.dists = NULL, edge.density = 0.5, verbose = FALSE)

Arguments

node.name	a vector of character giving the names of the nodes. It gives the size of the simulated DAG.
data.dists	named list corresponding to the node.name specifying the distribution for each node. If not provided arbitrary distributions are assigned to the nodes.
edge.density	number in [0,1] specifying the edge probability in the dag.
verbose	print more information on the run.

Details

This function generates DAGs by sampling triangular matrices and reorder columns and rows randomly. The network density (edge.density) is used column-wise as binomial sampling probability. Then the matrix is named using the user-provided names.

Value

object of class abnDag consisting of a named matrix, a named list giving the distribution for each node and an empty element for the data.

Examples

simdag <- simulateDag(node.name = c("a", "b", "c", "d"),
                      edge.density = 0.5,
                      data.dists = list(a = "gaussian",
                                        b = "binomial",
                                        c = "poisson",
                                        d = "multinomial"))

## Example using Ozon entries:
dist <- list(Ozone="gaussian",   Solar.R="gaussian",  Wind="gaussian",
             Temp="gaussian",    Month="gaussian",    Day="gaussian")
out <- simulateDag(node.name = names(dist), data.dists = dist, edge.density = 0.8)
plot(out)

abn documentation built on Nov. 3, 2023, 5:08 p.m.