@incollection{MIP-0905,
author={M. T. Gallegos, G. Ritter},
title={{Trimming algorithms for clustering contaminated grouped data and their robustness}},
institution={{Fakult{\"a}t f{\"u}r Informatik und Mathematik, Universit{\"a}t Passau}},
year={2009},
number={MIP-0905}
}
We establish an affine equivariant, constrained heteroscedastic model and criterion with trimming for clustering contaminated, grouped data. We show existence of the m.l.e., propose a method for determining an appropriate constraint, and design a strategy for finding reasonable partitions. We finally compute breakdown points of the estimated parameters thereby showing asymptotic robustness of the method.
