networkAnalysis: Estimate network structures

In SachaEpskamp/psynet: Automatic network analysis of psychological data

Usage

networkAnalysis(x, methods = c("cor", "pcor", "alpcor", "plspcor", "pc", "pcskel", "BDbest", "BDpost", "BDavg", "bn.gs", "bnboot.gs", "bn.iamb", "bnboot.iamb", "bn.hc", "bnboot.hc", "bn.tabu", "bnboot.tabu", "bn.mmhc", "bnboot.mmhc", "bn.rsmax2", "bnboot.rsmax2", "bn.mmpc", "bnboot.mmpc", "bn.si.hiton.pc", "bnboot.si.hiton.pc", "bn.chow.liu", "bnboot.chow.liu", "bn.aracne", "bnboot.aracne", "IsingFit"), scale, nonparanormal = FALSE, ask = FALSE, titles = TRUE, citations = TRUE, layout = "spring", layoutToFirst = TRUE, pcAlpha = 0.05, BDargs = list(), bnlearnArgs = list(), bnbootArgs = list(R = 200), parallelEdge = TRUE, labels = TRUE, verbose = TRUE, pcNoDicho = FALSE, adaptDF = TRUE, graphArgs = list(), IsingFitArgs = list(), ...)

Arguments

x	A dataset with rows indicating measures and columns indicating variables. All variables must be of the same scale: dichotomous, ordinal or continous.
methods	A vector of strings indicating which networks to estimate. The function will estimate all graphs and plot them after each other (e.g., so that pdf() will create a page for each plot) with layout based on the first graph (rearainging the order of this argument causes the layout to be based on a different graph). Currently supported are: "cor", "pcor", "alpcor", "plspcor", "IsingFit", "pc", "pcskel", "BDbest", "BDpost", "BDavg", "bn.gs", "bnboot.gs", "bn.iamb", "bnboot.iamb", "bn.hc", "bnboot.hc", "bn.tabu", "bnboot.tabu", "bn.mmhc", "bnboot.mmhc", "bn.rsmax2", "bnboot.rsmax2", "bn.mmpc", "bnboot.mmpc", "bn.si.hiton.pc", "bnboot.si.hiton.pc", "bn.chow.liu", "bnboot.chow.liu", "bn.aracne" and "bnboot.aracne"
scale	"dichotomous", "ordinal" or "continuous". Is otherwise detected.
nonparanormal	Should the nonparanormal transfromation from huge.npn be used on the data?
ask	Logical indicating if R should ask to go to the next plot.
titles	Logical indicating if titles should be added.
citations	Logical indicating if citations should be added.
layout	Layout to be used in all graphs
layoutToFirst	Equate layout to the layout of the first plot?
pcAlpha	Alpha used in pcalg functions
BDargs	List of arguments used in bdgraph
bnlearnArgs	List of arguments used in bn
bnbootArgs	List of arguments used in bn.boot
parallelEdge	Same as in qgraph
labels	Same as in qgraph
verbose	Should messages be printed to the console indicating what is happening?
pcNoDicho	Logical. If TRUE pcalg for dichotomous data will be run on tetrachoric correlations and not the discrete version. Seems to have more power.
adaptDF	Argument used in pcalg
graphArgs	Named list with qgraph args that can be used to specify specific arguments. E.g., list(alpcor = list(fade = FALSE))
IsingFitArgs	Arguments used in IsingFit
...	Arguments sent to qgraph

Value

A named list with for each estimated graph:

graph	The weights matrix of the graph
output	Output of the method
qgraph	The qgraph object
method	A string indicating the method used

Author(s)

Sacha Epskamp <[e-mail]>

Examples

# Load smoking dataset:
data(smoking)

# Estimate three networks:
Res <- networkAnalysis(smoking, methods = c( "pc", "IsingFit" , "bn.gs"), ask = FALSE)

# Centrality measures:
centralityPlot(Res)

# Clustering:
centralityPlot(Res)

SachaEpskamp/psynet documentation built on May 9, 2019, 12:09 p.m.