easybgm (version 0.2.1)

sparse_or_dense: Test for sparse against dense topologies

Description

The function tests if a network is sparse (i.e., few edges in between nodes) or dense (i.e., a lot of edges between nodes). It estimates the network model of a given data set under the hypothesis that it is sparse and that it is dense, and computes th Bayes factor.

Usage

sparse_or_dense(x, type, ...)

Value

List containing results of the analysis:

• log.BF The log of the Bayes factor of the test of a sparse against a dense network structure. A value larger 0 indicates evidence for a sparse structure. Contrarily, a value smaller 0 indicates evidence for a dense structure.

• BF The Bayes factor of the test of a sparse against a dense network structure. A value larger 1 indicates evidence for a sparse structure. Contrarily, a value between 0 and 1 indicates evidence for a dense structure.

• relative.complexity.sparse The relative complexity under a sparse prior hypothesis, which is the proportion of estimated included edges relative to the total possible edges under the different hypotheses.

• relative.complexity.dense The relative complexity under a dense prior hypothesis.

• relative.complexity.uniform The relative complexity under a uniform prior hypothesis.

• no.hypotheses The number of hypotheses that are computed in the analysis. At least the sparse, dense and uniform are used, but sometimes additional bridge hypotheses have to be computed to be able to determine the Bayes factor.

Arguments

x

An n x p matrix or dataframe containing the variables for n independent observations on p variables.

type

What is the data type? Options: currently only binary and ordinal are supported

...

additional arguments of the bgms function

Examples

# \donttest{
library(easybgm)
library(bgms)

data <- na.omit(Wenchuan)

# Fitting the Wenchuan PTSD data

fit <- sparse_or_dense(data, type = "ordinal",
                iter = 1000 # for demonstration only (> 5e4 recommended)
                )
# }