calculates variable and variable-cluster importances

Share:

Description

The function computes the log Bayes factors for the hypothesis H0: the variable or the variable-cluster combination is useful for clustering against H1: the variable or the variable-cluster combination is useless. The Bayes factors are computed for the optimal allocation found by the bclust function.

Usage

1
imp(x)

Arguments

x

A bclustvs object.

Value

var

A vector being the log Bayes factor of d_{v}=1 against d_{v}=0, see bclust for details.

varclust

A vector being the log Bayes factor of g_{vc}=1 against g_{vc}=0, see bclust for details.

repno

The number of replicates producing each row of varclust.

labels

The vector of variable labels extracted from the bclustvs object.

order

The order of var useful to sort var, varclust, and labels.

See Also

bclust.

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
data(gaelle)

gaelle.id<-rep(1:14,c(3,rep(4,13))) 
# first 3 rows replication of ColWT, 4 for the rest

gaelle.bclust<-bclust(gaelle,rep.id=gaelle.id,
transformed.par=c(-1.84,-0.99,1.63,0.08,-0.16,-1.68),
var.select=TRUE)

gaelle.imp<-imp(gaelle.bclust)

#plot the variable importances 
par(mfrow=c(1,1)) #retreive graphic defaults 

mycolor<-gaelle.imp$var
mycolor<-c()
mycolor[gaelle.imp$var>0]<-"black"
mycolor[gaelle.imp$var<=0]<-"white"

viplot(var=gaelle.imp$var,xlab=gaelle.imp$labels,col=mycolor)
#plot important variables with balck

viplot(var=gaelle.imp$var,xlab=gaelle.imp$labels,
sort=TRUE,col=heat.colors(length(gaelle.imp$var)),
xlab.mar=10,ylab.mar=4)
mtext(1, text = "Metabolite", line = 7,cex=1.5)# add x axis label
mtext(2, text = "Log Bayes Factor", line = 3,cex=1.2)# add y axis labels
#sort importnaces and use heat colors add some labels to the x and y axes