ueda

An R package with useful functions to automatically detect and remove outliers via the Ueda’s method. For more information see

Marmolejo-Ramos, F., Vélez, J.I. & Romão, X. Automatic detection of discordant outliers via the Ueda’s method. J Stat Distrib App 2, 8 (2015). https://doi.org/10.1186/s40488-015-0031-y

You can visit the package website to explore the vignettes (articles) and function reference.

Installation

To install the latest development version of the package from GitHub use:

## download and install necessary components
if(!require('devtools')) install.packages('devtools')
require('devtools')

if(!require('here')) install.packages('here')
require('here')

devtools::install_github('jivelez/ueda', force = TRUE)
require(ueda)

Examples

Example 1

First we generate a numeric vector x as follows:

# random seed for reproducibility
set.seed(13)

# generate data and introduce 5 outliers
x <- c(rnorm(25, 300, 10), rnorm(5, 400, 5))

# shuffle the data
x <- sample(x)
x

To get the full output after applying Ueda’s method on this data, we use the find_ueda_outliers() function. In this particular case, we use smax = 5 to detect up to 5 outliers in the numeric vector x:

# run full Ueda's method and detect up to 5 outliers
out <- find_ueda_outliers(x, smax = 5)
out

Note that out is a list of four elements: Ut, label, x and x_new. For more information type ?find_ueda_outliers in the R console.

Example 2

In this example, we use the numeric vector x as input and automatically detect up to smax = 5 outliers and remove them with the ueda_remove() function:

# remove up to 5 outliers from vector `x`
x_new <- ueda_remove(x, smax = 5)
x_new

Note that the x_new object is similar to x but has no outliers. For more information type ?ueda_remove in the R console.

Example 3

Suppose we have the following data.frame():

## create some data
set.seed(721)
y <- sample(0:1, length(x), replace = TRUE)
d <- data.frame(x = x, y = y)
d

Sometimes we would like to get the indexes of the outliers in the vector x. These indexes can be further used in combination with the subset() function to reduce the number of registers in a d.

One way of accomplishing this is using the udea_index() function as follows:

# get indexes and positions of up to 5 outliers
result <- ueda_index(x, smax = 5)
result

The object result is a list() of two components: indexes and positions. For more information type ?ueda_index in the R console.

To get the indexes from the object result we proceed as follows:

## indexes of outliers
idx <- result$indexes
idx

The object idx can be used to subset() the object d. To get the complete data set for the outlier observations, we do

## data for outlier observations
subset(d, idx)

The data for those observations that are not considered outliers can be retrieved by doing

## data for no outlier observations
subset(d, !idx)