outliers_mad: Identify outliers using robust median absolute deviation...

In hauselin/hausekeep: Miscellaneous helper and utility functions

Identify outliers using robust median absolute deviation approach

Description

outliers_mad is used to identify outliers in vectors using Leys et al.'s (2003) median absolute deviation approach.

Usage

outliers_mad(x, threshold = 3.0, replace_outlier_value = NA,
show_mad_values = FALSE, show_outlier_indices = FALSE,
b_constant = 1.4826, digits = 2, debug = FALSE)

Arguments

x	a vector of numbers
threshold	value to use as cutoff (Leys et al. recommend 2.5 or 3.0 as default)
replace_outlier_value	if value is an outlier, what to replace it with? NA by default
show_mad_values	if TRUE, will show deviation score of each value
show_outlier_indices	if TRUE, return index/position of outliers
b_constant	a constant linked to the assumption of normality of the data, disregarding the abnormality induced by outliers
digits	how many digits to round output to
debug	if TRUE, print messages (FALSE by default)

Details

We can identify and remove outliers in our data by identifying data points that are too extreme—either too many standard deviations (SD) away from the mean or too many median absolute deviations (MAD) away from the median. The SD approach might not be ideal with extreme outliers, whereas the MAD approach is much more robust (for comparison of both approaches, see Leys et al., 2013, Journal of Experimental Social Psychology).

b_constant is usually 1.4826, a constant linked to the assumption of normality of the data, disregarding the abnormality induced by outliers (Rousseeuw & Croux, 1993).

Value

A vector with outliers identified (default converts outliers to NA)

Author(s)

Hause Lin

References

• Leys, C., Ley, C., Klein, O., Bernard, P., & Licata, L. (2013). Detecting outliers: Do not use standard deviation around the mean, use absolute deviation around the median. Journal of Experimental Social Psychology, 49(4), 764-766. doi:10.1016/j.jesp.2013.03.013 (https://www.sciencedirect.com/science/article/pii/S0022103113000668)

See Also

outliersZ

Examples

x <- c(1, 3, 3, 6, 8, 10, 10, 1000, -1000) # 1000 is an outlier
outliers_mad(x)
outliers_mad(x, threshold = 3.0)
outliers_mad(x, threshold = 2.5, replace_outlier_value = -999)
outliers_mad(x, threshold = 1.5, show_outlier_indices = TRUE)
outliers_mad(x, threshold = 1.5, show_mad_values = TRUE)
outliers_mad(x, threshold = 1.5, show_mad_values = TRUE, replace_outlier_value = -88)

hauselin/hausekeep documentation built on Feb. 3, 2023, 3:09 p.m.