pdag2dag: Extend a Partially Directed Acyclic Graph (PDAG) to a DAG

In pcalg: Methods for Graphical Models and Causal Inference

Extend a Partially Directed Acyclic Graph (PDAG) to a DAG

Description

This function extends a PDAG (Partially Directed Acyclic Graph) to a DAG, if this is possible.

Usage

pdag2dag(g, keepVstruct=TRUE)

Arguments

g	Input PDAG (graph object)
keepVstruct	Logical indicating if the v-structures in g are kept. Otherwise they are ignored and an arbitrary extension is generated.

Details

Direct undirected edges without creating directed cycles or additional v-structures. The PDAG is consistently extended to a DAG using the algorithm by Dor and Tarsi (1992). If no extension is possible, a DAG corresponding to the skeleton of the PDAG is generated and a warning message is produced.

Value

List with entries

graph	Contains a consistent DAG extension (graph object),
success	Is TRUE iff the extension was possible.

Author(s)

Markus Kalisch [e-mail]

References

D.Dor, M.Tarsi (1992). A simple algorithm to construct a consistent extension of a partially oriented graph. Technicial Report R-185, Cognitive Systems Laboratory, UCLA

Examples

p <- 10 # number of random variables
n <- 10000 # number of samples
s <- 0.4 # sparsness of the graph

## generate random data
set.seed(42)
g <- randomDAG(p, prob = s) # generate a random DAG
d <- rmvDAG(n,g) # generate random samples

gSkel <- pcAlgo(d,alpha=0.05) # estimate of the skeleton

(gPDAG <- udag2pdag(gSkel))

(gDAG <- pdag2dag(gPDAG@graph))

pcalg documentation built on May 29, 2024, 5:24 a.m.