baysout: Outlier detection using Bay and Schwabacher's algorithm.

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

Description

This function implements the algorithm for outlier detection found in Bay and Schwabacher(2003). The algorithm assigns an outlyingness measure to each observation and returns the indexes of those observations having the largest measures. The number of outliers to be returned is specified by the user.

Usage

baysout(D, blocks = 10, nclass=0, k = 3, num.out = 10)

Arguments

D	the dataset under study
blocks	the number of sections in which to divide the entire dataset. It must be at least as large as the number of outliers requested.
nclass	To find the outliers without taking in cnsideration the feature class enter 0. To find the outliers for a given class enter the class number.
k	the number of neighbors to find for each observation
num.out	the number of outliers to return

Value

num.out

Returns a two column matrix containing the indexes of the observations with the top num.out outlyingness measures. A plot of the top candidates and their measures is also displayed.

Author(s)

Caroline Rodriguez(2004). Modified by Elio Lozano (2005) and Edgar Acuna (2015)

References

Bay, S.D., and Schwabacher (2003). Mining distance-based outliers in near linear time with randomization and a simple pruning rule.

Examples

#---- Outliers detection using the Bay's algorithm----
data(bupa)
bupa.out=baysout(bupa[bupa[,7]==1,1:6],blocks=10,num.out=10)

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