svm_margin_filter: Filter points using SVM hyperplane

In camroach87/semisupervisr: Semi-supervised training functions

Description

Filters points using SVM hyperplane and margin. Four regions are considered from which rho points are returned.

Usage

svm_margin_filter(object, data, x, rho)

Arguments

object	Object of class "svm".
data	Data frame to be filtered.
x	Feature columns.
rho	Number of points to be returned from each

Details

The four regions of the sample space are,

1. Points above the hyperplane and within the margin. Points closest to the margin boundary are returned.

2. Points below the hyperplane and within the margin. Points closest to the margin boundary are returned.

3. Points above the hyperplane. Points furthest from the hyperplane are returned.

4. Points below the hyperplane. Points furthest from the hyperplane are returned.

If there are less than rho points remaining in any of these spaces only the remaining points will be returned.

Value

A data frame of filtered values that can be added to the labelled data set.

Author(s)

Cameron Roach

camroach87/semisupervisr documentation built on May 13, 2019, 11:04 a.m.