R Graphical Manual

Browse All

Last data update: 2014.03.03

R: Outlier detection

outlier

R Documentation

Outlier detection

Description

Identifies outliers based on the nearest neighbour criterion. It starts by computing a matrix of distances (correlation, r, used as distance, dr=(1-r)/2). Variables with nearest neighbour distance larger than parameter thresh are considered outliers.

Usage

outlier(veg, thresh, y,...)
outly(veg, thresh = 0.2, y = 0.5)

## Default S3 method:
outlier(veg, thresh, y,...)
## S3 method for class 'outlier'
plot(x,...)
## S3 method for class 'outlier'
print(x,...)

Arguments

`veg`	This is a vegetation data frame, releves are rows, species columns
`thresh`	Threshold nearest neighbour distance for outliers
`y`	Transformation of species scores: x'= x exp(y)
`x`	An object of class "outlier"
`...`	Parameter out.seq, the plotting interval

Value

An object of class "oulier" with at least the following items:

`threshold`	Threshold nearest neighbour distance for considering outliers
`y`	Transformation of species scores: x'= x exp(y)
`rel.names`	All row names
`neigh.names`	Names of the corresponding nearest neighbours
`neigh.dist`	Distance to the nearest neighbour
`olddim`	Dimensions of data frame veg
`newdim`	Dimensions of data frame with outliers erased
`new.data`	Vegetation data frame without outliers
`pco.points`	The pco ordination scores use for plotting

Author(s)

Otto Wildi

References

Wildi, O. 2013. Data Analysis in Vegetation Ecology. 2nd ed. Wiley-Blackwell, Chichester.

Examples

o.outlier<- outlier(nveg,thresh=0.2,y=0.5)
o.outlier                                    # a list of all variables
plot(o.outlier)                              # nearest neighbour histogram and
                                             # pco ordination