thin_edges: Thinning of edges in a graphical model with a triangulated...

In nvihrs14/tcherry: Learning the structure of tcherry trees

Description

Thinning of edges in a graphical model with a triangulated graph based on a likelihood ratio test for conditional independence.

Usage

thin_edges(cliques, separators, data, alpha = 0.05, ...)

Arguments

cliques	A list containing the cliques of the triangulated graph, which should be thinned.
separators	A list containing the separators of the junction tree of the triangulated graph, which should be thinned.
data	Data with realisations of the variables in the graph.
alpha	Significance level used in the conditional independece test.
...	Additional arguments passed to cond_independence_test

Details

The edges in the graph are removed one by one if the respective conditional independence test cannot be rejected. An edge is only considered if its removal results in a new triangulated graph, i.e. the edge is in one clique only.

Value

A list containing the following components:

• adj_matrix The adjacency matrix of the thinned graph.

• cliques The cliques of the thinned graph.

• separators The separators of the junction tree for the thinned graph.

• n_edges The number of edges in the resulting graph.

• n_edges_removed The number of removed edges.

Author(s)

Katrine Kirkeby, [e-mail]

Maria Knudsen, [e-mail]

Ninna Vihrs, [e-mail]

See Also

cond_independence_test for the test used.

Examples

set.seed(43)
var1 <- c(sample(c(1, 2), 100, replace = TRUE))
var2 <- var1 + c(sample(c(1, 2), 100, replace = TRUE))
var3 <- var1 + c(sample(c(0, 1), 100, replace = TRUE,
                        prob = c(0.9, 0.1)))
var4 <- c(sample(c(1, 2), 100, replace = TRUE))
var5 <- var2 + var3
var6 <- var1 - var4 + c(sample(c(1, 2), 100, replace = TRUE))
var7 <- c(sample(c(1, 2), 100, replace = TRUE))

data <- data.frame("var1" = as.character(var1),
                   "var2" = as.character(var2),
                   "var3" = as.character(var3),
                   "var4" = as.character(var4),
                   "var5" = as.character(var5),
                   "var6" = as.character(var6),
                   "var7" = as.character(var7))

cliques <- list(c("var1", "var2", "var3"),
                c("var2", "var3", "var5"),
                c("var5", "var6", "var7"),
                c("var1", "var4"),
                c("var2", "var5", "var6"))

separators <- list(c("var1"),
                   c("var2", "var3"),
                   c("var2", "var5"),
                   c("var5", "var6"))

thin_edges(cliques, separators, data = data, alpha = 0.1,
               smooth = 0.1)

nvihrs14/tcherry documentation built on Aug. 1, 2020, 6:25 p.m.