bnlearn - man/learn.html

single-node local discovery {bnlearn}

R Documentation

Discover the structure around a single node

Description

Learn the Markov blanket or the neighbourhood centered on a node.

Usage

learn.mb(x, node, method, whitelist = NULL, blacklist = NULL, start = NULL,
  test = NULL, alpha = 0.05, ..., max.sx = NULL, debug = FALSE)
learn.nbr(x, node, method, whitelist = NULL, blacklist = NULL,
  test = NULL, alpha = 0.05, ..., max.sx = NULL, debug = FALSE)

Arguments

`x`	a data frame containing the variables in the model.
`node`	a character string, the label of the node whose local structure is being learned.
`method`	a character string, the label of a structure learning algorithm. Possible choices are listed in structure learning.
`whitelist`	a vector of character strings, the labels of the whitelisted nodes.
`blacklist`	a vector of character strings, the labels of the blacklisted nodes.
`start`	a vector of character strings, the labels of the nodes to be included in the Markov blanket before the learning process (in `learn.mb`). Note that the nodes in `start` can be removed from the Markov blanket by the learning algorithm, unlike the nodes included due to whitelisting.
`test`	a character string, the label of the conditional independence test to be used in the algorithm. If none is specified, the default test statistic is the mutual information for categorical variables, the Jonckheere-Terpstra test for ordered factors and the linear correlation for continuous variables. See `independence tests` for details.
`alpha`	a numeric value, the target nominal type I error rate.
`...`	optional arguments to be passed to the test specified by `test`. See `ci.test` for details.
`max.sx`	a positive integer, the maximum allowed size of the conditioning sets used in conditional independence tests. The default is that there is no limit on size.
`debug`	a boolean value. If `TRUE` a lot of debugging output is printed; otherwise the function is completely silent.

Value

A vector of character strings, the labels of the nodes in the Markov blanket (for learn.mb()) or in the neighbourhood (for learn.nbr()).

Note

All algorithms used by learn.mb() and learn.nbr() accept incomplete data, which they handle by computing individual conditional independence tests on locally complete observations.

Author(s)

Marco Scutari

Examples

learn.mb(learning.test, node = "D", method = "iamb")
learn.mb(learning.test, node = "D", method = "iamb", blacklist = c("A", "F"))

learn.nbr(gaussian.test, node = "F", method = "si.hiton.pc", whitelist = "D")

Index

Topics