adjustment: Compute adjustment sets for covariate adjustment.

In pcalg: Methods for Graphical Models and Causal Inference

Compute adjustment sets for covariate adjustment.

Description

This function is a wrapper for convenience to the function adjustmentSet from package dagitty.

Usage

adjustment(amat, amat.type, x, y, set.type)

Arguments

amat	adjacency matrix of type amat.cpdag or amat.pag.
amat.type	string specifying the type of graph of the adjacency matrix amat. It can be a DAG (type="dag"), a CPDAG (type="cpdag") or a maximally oriented PDAG (type="pdag") from Meek (1995); then the type of adjacency matrix is assumed to be amat.cpdag. It can also be a MAG (type = "mag") or a PAG (type="pag"); then the type of the adjacency matrix is assumed to be amat.pag.
x	(integer) position of variable x in the adjacency matrix.
y	(integer) position of variable y in the adjacency matrix.
set.type	string specifying the type of adjustment set that should be computed. It can be "minimal" ,"all" and "canonical". See Details explanations.

Details

If set.type is "minimal", then only minimal sufficient adjustment sets are returned. If set.type is "all", all valid adjustment sets are returned. If set.type is "canonical", a single adjustment set is returned that consists of all (possible) ancestors of x and y, minus (possible) descendants of nodes on proper causal paths. This canonical adjustment set is always valid if any valid set exists at all.

Value

If adjustment sets exist, list of length at least one (list elements might be empty vectors, if the empty set is an adjustment set). If no adjustment set exists, an empty list is returned.

Author(s)

Emilija Perkovic and Markus Kalisch ([e-mail])

References

E. Perkovic, J. Textor, M. Kalisch and M.H. Maathuis (2015). A Complete Generalized Adjustment Criterion. In Proceedings of UAI 2015.

E. Perkovic, J. Textor, M. Kalisch and M.H. Maathuis (2017). Complete graphical characterization and construction of adjustment sets in Markov equivalence classes of ancestral graphs. To appear in Journal of Machine Learning Research.

B. van der Zander, M. Liskiewicz and J. Textor (2014). Constructing separators and adjustment sets in ancestral graphs. In Proceedings of UAI 2014.

See Also

gac for testing if a set satisfies the Generalized Adjustment Criterion.

Examples

## Example 4.1 in Perkovic et. al (2015), Example 2 in Perkovic et. al (2017)
mFig1 <- matrix(c(0,1,1,0,0,0, 1,0,1,1,1,0, 0,0,0,0,0,1,
                  0,1,1,0,1,1, 0,1,0,1,0,1, 0,0,0,0,0,0), 6,6)
type <- "cpdag"
x <- 3; y <- 6
## plot(as(t(mFig1), "graphNEL"))

## all
if(requireNamespace("dagitty", quietly = TRUE)) {
adjustment(amat = mFig1, amat.type = type, x = x, y = y, set.type =
"all")
}
## finds adjustment sets: (2,4), (1,2,4), (4,5), (1,4,5), (2,4,5), (1,2,4,5)

## minimal
if(requireNamespace("dagitty", quietly = TRUE)) {
adjustment(amat = mFig1, amat.type = type, x = x, y = y, set.type =
"minimal")
}
## finds adjustment sets: (2,4), (4,5), i.e., the valid sets with the fewest elements

## canonical
if(requireNamespace("dagitty", quietly = TRUE)) {
adjustment(amat = mFig1, amat.type = type, x = x, y = y, set.type =
"canonical")
}
## finds adjustment set: (1,2,4,5)

pcalg documentation built on Sept. 26, 2023, 9:06 a.m.