This function is DEPRECATED! Use skeleton, pc or
fci instead.
Use the PC-algorithm to estimate the underlying graph (“skeleton”) or the equivalence class (CPDAG) of a DAG.
pcAlgo(dm = NA, C = NA, n=NA, alpha, corMethod = "standard",
verbose=FALSE, directed=FALSE, G=NULL, datatype = "continuous",
NAdelete=TRUE, m.max=Inf, u2pd = "rand", psepset=FALSE)
Data matrix; rows correspond to samples, cols correspond to nodes.
Correlation matrix; this is an alternative for specifying the data matrix.
Sample size; this is only needed if the data matrix is not provided.
Significance level for the individual partial correlation tests.
A character string speciyfing the method for
(partial) correlation estimation.
"standard", "QnStable", "Qn" or "ogkQn" for standard and robust (based on
the Qn scale estimator without and with OGK) correlation
estimation. For robust estimation, we recommend "QnStable".
0-no output, 1-small output, 2-details;using 1 and 2 makes the function very much slower
If FALSE, the underlying skeleton is computed;
if TRUE, the underlying CPDAG is computed
The adjacency matrix of the graph from which the algorithm should start (logical)
Distinguish between discrete and continuous data
Delete edge if pval=NA (for discrete data)
Maximal size of conditioning set
Function used for converting skeleton to cpdag. "rand" (use udag2pdag); "relaxed" (use udag2pdagRelaxed); "retry" (use udag2pdagSpecial)
If true, also possible separation sets are tested.
An object of class "pcAlgo" (see
'>pcAlgo) containing an undirected graph
(object of class "graph", see
graph-class from the package graph)
(without weigths) as estimate of the skeleton or the CPDAG of the
underlying DAG.
P. Spirtes, C. Glymour and R. Scheines (2000) Causation, Prediction, and Search, 2nd edition, The MIT Press.
Kalisch M. and P. B\"uhlmann (2007) Estimating high-dimensional directed acyclic graphs with the PC-algorithm; JMLR, Vol. 8, 613-636, 2007.