hist_out: Outlier detection (histogram)

In privefl/bigutilsr: Utility Functions for Large-scale Data

Outlier detection (histogram)

Description

Outlier detection based on departure from histogram. Suitable for compact values (need a space between main values and outliers).

Usage

hist_out(x, breaks = nclass.scottRob, pmax_out = 0.2, nboot = NULL)

Arguments

x	Numeric vector (with compact values).
breaks	Same parameter as for hist(). Default uses a robust version of Scott's rule. You can also use "FD" or nclass.FD for a bit more bins.
pmax_out	Percentage at each side that can be considered outliers at each step. Default is 0.2.
nboot	Number of bootstrap replicates to estimate limits more robustly. Default is NULL (no bootstrap, even if I would recommend to use it).

Value

A list with

• x: the initial vector, whose outliers have been removed,

• lim: lower and upper limits for outlier removal,

• all_lim: all bootstrap replicates for lim (if nboot not NULL).

Examples

set.seed(1)
x <- rnorm(1000)
str(hist_out(x))

# Easy to separate
x2 <- c(x, rnorm(50, mean = 7))
hist(x2, breaks = nclass.scottRob)
str(hist_out(x2))

# More difficult to separate
x3 <- c(x, rnorm(50, mean = 6))
hist(x3, breaks = nclass.scottRob)
str(hist_out(x3))
str(hist_out(x3, nboot = 999))

privefl/bigutilsr documentation built on Jan. 27, 2024, 5:33 a.m.