confusion: Confusion matrix

In jmbh/mMRF: Estimating mixed Markov random fields

Description

Computes accuracy, sensitivity, specificity and precision from a true and an estimated adjacency matrix.

Usage

confusion(tg, eg)

Arguments

tg	p x p adjacenty matrix of the true graph
eg	p x p adjacenty matrix of the estimated graph

Value

A list containing accuracy, sensitivity, specificity and precision of the estimated graph in reference to the true graph.

Author(s)

Jonas Haslbeck

Examples

#create some data from a mixed graphical model
n <- 100 # number of samples
type <- c("g", "c", "c") # one gaussian, two categorical
lev <- c(1,3, 2) # the categorical variable have 3 and 2 categories, respectively
graph <- matrix(0,3,3) 
graph[1,2] <- graph[2,1] <- .5 # we have an edge with weight .5 between node 1 and 2
thresh <- list(c(0), c(0,0,0), c(0,0)) # all thresholds are zero (for the categorical variables each categoriy has a threshold)
data <- mMRFsampler(n, type, lev, graph, thresh, parmatrix=NA, nIter=1000)

#fit a mixed graphical model
fit <- mMRFfit(data, type, lev=lev, d=2)
grapht <- graph
grapht[grapht!=0] <- 1 #binarize true graph

confusion(grapht, fit$adj) #compute derivatives of confusion matrix

jmbh/mMRF documentation built on May 19, 2019, 1:51 p.m.