Title: TCLUST extensions
Authors: Agustin Mayo-Iscar - Universidad de Valladolid (Spain) [presenting]
Luis Angel Garcia-Escudero - Universidad de Valladolid (Spain)
Alfonso Gordaliza - Universidad de Valladolid (Spain)
Francesca Greselin - University of Milano Bicocca (Italy)
Abstract: TCLUST is a model-based clustering robust methodology for multivariate data. Its robustness is based in the joint application of trimming and constraints. TCLUST extensions have been developed for estimating mixtures of linear models, mixtures of factor analyzers and mixtures of Skew-Normal models. In every of these frameworks, this methodology shows good performance when applied to data sets containing contaminated observations. Data driven tools for helping to the users in choosing the input parameters, related with the joint application of these tools, are also available.