outliers detection functions

Description

Distribution based outlier detection functions.

qoutlier is IQR based outlier detection.

outlier.norm is based on normal distribution using Huber M-estimator of location with MAD scale

outlier.t is based on t-distribution.

outlier.cutoff is a simple cutoff-based outlier detection.

Usage

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
proportion.outliers.robust(x, alpha = 0.01, isUpper = TRUE,
  isLower = TRUE)

proportion.outliers.mle(x, alpha = 0.01, isUpper = TRUE, isLower = TRUE)

qoutlier(x, alpha = 1.5, isUpper = TRUE, isLower = TRUE, plot = FALSE,
  ...)

outlier.norm(x, alpha = 0.01, z.cutoff = NULL, isUpper = TRUE,
  isLower = TRUE, plot = FALSE)

outlier.t(x, alpha = 0.01, z.cutoff = NULL, isUpper = TRUE,
  isLower = TRUE, plot = FALSE)

outlier.cutoff(x, lBound = NULL, uBound = NULL)

Arguments

x

An integer/numeric vector used as the input

alpha,z.cutoff

alpha is the percentage of the standard deviation from the center of the data. z.cutoff is the standardized z-score value. They are used as the distribution based thresholds.

isUpper,isLower

logical scalars indicating whether the outliers are checked at upper or lower side of the distribution.

plot

logical scalar indicating whether to visualize the outlier detection results.

...

other arguments to be passed to qoutlier function,currently it is ignored.

lBound,uBound

Numeric scalars used as cutoff threshold for either lower limit or upper limit

Details

These different outlier detection functions are used together with qaCheck method to perform outlier checks.

Value

a logical vector with the same length of input vector,indicating whether each entry of the input is a outlier.

Author(s)

Mike Jiang,Greg Finak

Maintainer: Mike Jiang <wjiang2@fhcrc.org>

See Also

qaCheck,qaReport