ChooseMetaVars {rWMBAT}R Documentation

ChooseMetaVars Attempts To Combine Variables into Better Variables

Description

Finds the V-V pairs in the adjacency matrix, and attempts to combine them into a metavariable with a higher mutual information than either variable alone. If it is possible to do this, it returns a new data matrix with the variables combined.

Usage

ChooseMetaVars(data, class, adj)

Arguments

data double array of discrete integer values, cases in rows and variables in columns.
class double column vector. Classification of each case
adj logical adjacency matrix, variables+1 by variables. Last row is class node. Logical meaning "there is an arc from i to j.

Value

metamatrix logical whose (i,j) means "variable j was combined into variable i (and erased)"
finaldata double data matrix with the variable combined and rebinned
leftbound vector,the new left boundary (vector) for binning
rightbound vector,the new right boundary (vector) for binning

Note

CALLED FUNCTIONS: opt3bin(rebins combined variables to determine highest MI)

Author(s)

Karl Kuschner, Qian Si and William Cooke, College of William and Mary, Dept. of Physics, 2009

References

http://kwkusc.people.wm.edu/dissertation/dissertation.htm

Examples

data(traingrp, traingrpclass, adjmat) # load example input data in the package
result <- ChooseMetaVars (traingrp, traingrpclass, adjmat)
         

[Package rWMBAT version 2.0 Index]