robout: Outlier Detection with Robust Mahalonobis distance

In dprep: Data Pre-Processing and Visualization Functions for Classification

Description

This function finds the outliers of a dataset using robust versions of the Mahalanobis distance.

Usage

robout(data, nclass=0, meth = c("mve", "mcd"), rep = 10, 
plot = TRUE)

Arguments

data	The dataset for which outlier detection will be carried out.
nclass	An integer value that represents the class to detect for outliers. By default nclass=0 meaning the column of classes it is not used.
meth	The method used to compute the Mahalanobis distance, "mve"=minimum volume estimator, "mcd"=minimum covariance determinant
rep	Number of repetitions
plot	A boolean value to turn on and off the scatter plot of the Mahalanobis distances

Details

It requires the use of the cov.rob function from the MASS library.

Value

top1	Index of observations identified as top outliers by frequency of selection
topout	Index of observations identified as possible outliers by outlyingness measure
outme	Index of observations and their outlyingness measures

Author(s)

Edgar Acuna

References

Rousseeuw, P, and Leroy, A. (1987). Robust Regression and outlier detection. John Wiley & Sons. New York.

Atkinson, A. (1994). Fast very robust methods for the detection of multiple outliers. Journal of the American Statistical Association, 89:1329-1339.

See Also

robout

Examples

## Not run: #---- Outlier Detection in bupa-class 1 using MCD
data(bupa)
robout(bupa,1,"mcd")

## End(Not run)

dprep documentation built on May 29, 2017, 11:01 a.m.