plotingPostOrder: Ploting post order

In andreamrau/GBNcausal: Causal estimation for Gaussian Bayesian networks

Description

This function plot the probability a node has to be on the ith position in the graph, according to the results of a MCMC.GBN run or a ObsOnly run.

The more dark the color is, the higher is the probability.

Usage

plotingPostOrder(W, refNames, obs = "", GBN.all)

Arguments

W	matrix - The true weightmatrix (direct causal effects) of the GBN.
refNames	vector - A vector of character, names of the nodes N# where # is a number.
obs	If obs = "obsOnly", it means the type of results is produced by observational data only.
GBN.all	list - A list of full.run (MCMC.GBN or ObsOnly results).

Value

It returns as many graphs as there is full.run in GBN.all list.

Examples

# Data creation

seed = 1990
n = 3000
p <- 10
m<-rep(0,10)
sigma<-rep(0.1,10) 

W <- 1*upper.tri(matrix(0,p,p))

data <- dataCreate(nbData = 2*p, p = 10,KO = list(1,9), nbKO = c(p,p), W = W , m = m,sigma = sigma, seed = seed)$data

# Initial Value

W1=1*upper.tri(matrix(0,p,p)) 
m1=rep(0,p)
s1=rep(10e-4,p)
colnames(W1)=names(m1)=names(s1)=rownames(W1)=paste("N",1:p,sep="")

firstGBN = new("GBNetwork",WeightMatrix=W1,resMean=m1,resSigma=s1)
firstGBN = GBNmle(firstGBN,data,lambda= 0,sigmapre=s1)$GBN

# Algorithm

results=MCMC.GBN(data, firstGBN, nbSimulation=2000, burnIn=20, seq=1, verbose=TRUE,verbose.index=100, 
              alpha=1,lambda=0) 
			  

refNames <- paste("N",1:10,sep = "")
plotingPostOrder(W1,refNames,obs = "",GBN.all = list(results))

andreamrau/GBNcausal documentation built on May 12, 2019, 3:34 a.m.