perturbation: Add Random Noise to Data

In shuklak13/anonymizeR: Simple Functions to Anonymize Data

Description

Add Random Noise to Data

Usage

perturbation(x, cols = 1:ncol(x), magnitude = "MEDIUM", minNoise = NULL,
  maxNoise = NULL, minNoiseQuantile = NULL, maxNoiseQuantile = NULL)

randomNoise(x, cols = 1:ncol(x), magnitude = "MEDIUM", minNoise = NULL,
  maxNoise = NULL, minNoiseQuantile = NULL, maxNoiseQuantile = NULL)

Arguments

x	a data.frame
cols	the columns that are to be affected. Must be numeric. If no value is given, all columns are affected.
magnitude	a shortcut for the potential size of noise - see details
minNoise	the smallest possible value for noise that may be added. If neither minNoise nor minNoiseQuantile is provided, minNoise defaults to 0.
maxNoise	the largest possible value for noise that may be added
minNoiseQuantile	the smallest possible value for noise that may be added. Rather than an actual value, this is a decimal, representing a quantile of the values in that column. If both minNoise and minNoiseQuantile are provided, maxNoise takes priority.
maxNoiseQuantile	the largest possible value for noise that may be added. Rather than an actual value, this is a decimal, representing a quantile of the values in that column. If both maxNoise and maxNoiseQuantile are provided, maxNoise takes priority.

Details

Adds a random value between minNoise and maxNoise to each element in each of the specified columns. If minNoiseQuantile or maxNoiseQuantile is provided instead of minNoise or maxNoise, then minNoise and maxNoise are calculated for every column based on the quantiles provided. magnitude is a shortcut for minNoise and maxNoiseQuantile parameters:

• LOW: minNoise = 0 maxNoiseQuantile = .25

• MEDIUM: minNoise = 0 maxNoiseQuantile = .5

• HIGH: minNoise = 0 maxNoiseQuantile = .75

This makes magnitude ideal when the user desires to have a noise value between 0 and some quantile for all specified columns.

Value

data.frame with with noise added all elements in specified columns

Functions

• randomNoise: identical to perturbation

Examples

perturbation(data)
perturbation(data, 1:5)
perturbation(data, magnitude="HIGH")
perturbation(data, "Gender", minNoise = 10, maxNoise = 50)
perturbation(data, minNoiseQuantile = .1, maxNoise = .9)

shuklak13/anonymizeR documentation built on May 29, 2019, 9:27 p.m.