ccd: Cyclic Causal Discovery Search

In bd2kccd/r-causal: R-Causal Library

Cyclic Causal Discovery Search

Description

ccd is the Cyclic Causal Discovery algorithm.

Usage

ccd(df, continuous = TRUE, depth = 3, significance = 0.05,
    verbose = FALSE, java.parameters = NULL, priorKnowledge = NULL)

Arguments

df	Data Frame of the dataset.
continuous	boolean indicating the dataset is continuous. continuous is TRUE by default.
depth	integer indicating a number of nodes conditioned on in the search. It ranges from -1 for unlimited nodes, otherwise a number >= 0. By default, depth=3.
significance	floating point indicating the statistical value (between zero and one) for the conditional independence test. By default, significance=0.05.
verbose	boolean indicating whether or not if verbose output should be printed. verbose is FALSE by default.
java.parameters	string indicating an optional parameters for JVM. For example, java.parameters = "-Xmx1024M". By default, java.parameters=NULL.
priorKnowledge	object indicating a prior knowledge of the graph. By default, priorKnowledge=NULL.

Details

The Cyclic Causal Discovery algorithm (CCD) is implemented as described by Thomas Richardson and Peter Spirtes in Chapter 7 of Computation, Causation, & Discovery by Glymour and Cooper eds. (pp. 269-271).

The search method returns an instance of a Graph but it also constructs two lists of node triples which represent the underlines and dotted underlines that the algorithm discovers.

More detail about ccd implementation, please visit the java code.

Value

A list containing the ccd's parameters, the input dataset, the result's nodes, and the result's edges.

Examples

data("audiology")
#Compute Cyclic Causal Discovery
ccd <- ccd(df=audiology, continuous = FALSE, depth = -1, verbose = TRUE)
ccd$parameters #Show the ccd's parameters
ccd$datasets #Show the dataset
ccd$nodes #Show the result's nodes
ccd$edges #Show the result's edges

bd2kccd/r-causal documentation built on Aug. 29, 2023, 9:17 a.m.