define_fixbox: Defining a "fixbox"-Object

In ao90/PRIM: Patient Rule Induction Method (PRIM)

Description

Function for determining on one box out of a "peel"-object.

Usage

define_fixbox(prim, step)

Arguments

prim	object of class "peel" from which the box should be defined.
step	number of the step (i.e. row in the argument box of the "peel"-object.) of the requested box.

Value

define_fixbox returns a object of class "fixbox", which is the basically same as a "peel"-object, but only contains one box. It is a list containing at least the following elements:

f	target function of the selected box.
beta	support of the selected box.
box	a data.frame with one row defining the borders of the box. The columns with "min." and "max." describe the lower and upper boundaries of the at least ordinal covariates. Therefore the value taken is the last one that is not included in the box. For the nominal variables there are columns for every category they can take. If the category is removed from the box the value FALSE is taken. The names of these columns are structured like: <variable name>.<category> For each variable with missing values (only if use_NAs = TRUE) there is also a column taking the value FALSE if the NAs of this variable are removed from the box. The names of these columns are structured like: <variable name>.NA
box_metric, box_nom, box_na	easier to handle definitions of the box for other functions
subset	logical vector indicating the subset (i.e. the observations that lie in the box).
data_orig	original dataset that was used for the peeling.

ao90/PRIM documentation built on May 5, 2019, 8:01 p.m.