REMOUTL

Remove outliers from a dataset

   B = REMOUTL(A,T,P)
   B = A*REMOUTL([],T,P)

Input
A	Dataset
T	Threshold for outlier detection (default 3)
P	Fraction of distances passing T (default 0.10)

Output
B	Dataset

Description

Outliers in A are removed, other objects are copied to B. Class by class  a distance matrix is constructed and objects are removed that have a fraction  P of their distances larger than the average distance in the class + T times  the standard deviation of the within-class distances. This routine works  also on unlabeled datasets. In partially labeled datasets the unlabeled  objects are neglected.