prune: Prune Points

In cccd: Class Cover Catch Digraphs

Prune Points

Description

a nearest neighbor pruning using neighborhood graphs.

Usage

prune(x, classes, prox = "Gabriel", ignore.ties = TRUE, ...)

Arguments

x	a data matrix.
classes	a vector of class labels.
prox	type of proximity graph.
ignore.ties	do not prune if there is a tie vote.
...	arguments passed to the proximity graph.

Details

First a proximity graph is computed on the data. Then points are marked if their neighbors have a different class than they do: if the most common class among the neighbors is different than the point. Then all marked points are removed.

Value

A list with attributes:

x	the pruned data.
v	the indices of the retained data.
g	the proximity graph.

Author(s)

David J. Marchette, [e-mail]

References

http://www.bic.mni.mcgill.ca/users/crisco/pgedit/

cccd documentation built on April 8, 2022, 5:08 p.m.