split.node {cubt}R Documentation

Computes the optimal split for a node using a criterion defined by crit.

Description

Computes the optimal split for a node using a criterion defined by crit.

Usage

 split.node(xx,N,crit0="anova")

Arguments

xx

Data matrix within the node to be split.

N

The whole sample size used to compute the criterion.

crit0

Choice for criterion: "anova" for using variance, "entropy" for entropy.

Value

Returns a list whose elements are used to construct the tree frame among whose the best split (coordinate - threshold) and the total deviance within the obtained children.

Author(s)

Badih Ghattas


[Package cubt version 3.3 Index]