Skip to contents

Summarize outlier values for numeric vectors

Source: R/summarize_outlier.R
summarize_outlier.Rd

Utilize check_outliers() to determine which values are determined as outliers. Summarizes the result to a comma-separated, decreasing string. Allows to specify a minimal amount of unique values in vec to perform outlier detection.

Usage

summarize_outlier(
  vec,
  minimal_unique_values = 10,
  method = c("iqr", "zscore_robust"),
  threshold = list(zscore_robust = 7, iqr = 5),
  sort = TRUE
)

Arguments

vec

The vector to be checked for outliers.

minimal_unique_values

Amount of unique values in vec before the outlier detection is performed. Set to 1 to always perform detection.

method

The outlier detection method(s). Can be "all" or some of "cook", "pareto", "zscore", "zscore_robust", "iqr", "ci", "eti", "hdi", "bci", "mahalanobis", "mahalanobis_robust", "mcd", "ics", "optics" or "lof".

threshold

A list containing the threshold values for each method (e.g. list('mahalanobis' = 7, 'cook' = 1)), above which an observation is considered as outlier. If NULL, default values will be used (see 'Details'). If a numeric value is given, it will be used as the threshold for any of the method run. For EFA/PCA/Omega, indicates the threshold for correlation of residuals (by default, 0.05).

sort

Sort the outlier values? Can be FALSE for no sorting, TRUE or "descending" for descending order, or "ascending" for ascending order.

Value

A summary string of outliers, an empty string if not outliers where detected.

Examples

norm <- rnorm(n = 100)
shift_norm <- rnorm(n = 10, mean = 999)
vec <- sample(c(norm, shift_norm))

summarize_outlier(vec)
#> [1] "1001.12685045903, 1001.0393692625, 1000.39181404564, 1000.07283825182, 999.44945377806, 999.426566546904, 999.424858441314, 999.249401783969, 999.107583992214, 997.31571846775"