join.cubt {cubt}R Documentation

Join non adjacent leaves of an unsupervised classification tree

Description

For each couple of leaves in a tree computes the gain of unifying them and join the couples who worth the best to be joint. The criterion used is the same as the one used to construct the tree (split the nodes)

Usage

 join.cubt(obj,Don,pas= NULL,nclass=4,verb=FALSE,crit0="anova") 

Arguments

obj

A cubt tree object.

Don

Data used to construct the tree

pas

Number of iterations done for the joining process. If not specified it equals the number of leaves.

nclass

Number of classes desired.

verb

Verbose.

crit0

Criterion used for joining, either "anova" either "entropy".

Value

A cubt tree where within each leaf the cluster number is given and not the number of observations.

Author(s)

Badih Ghattas


[Package cubt version 3.3 Index]