Robust estimation of mixture models with skew components via trimming and constraints

Impartial trimming procedures are commonly applied in many statistical settings for getting robust estimators in the presence of contamination. To get this robust behavior, when estimating mixture models, it is necessary to apply jointly trimming and constraints. Robust estimators based on these tools are available for estimating the model parameters in mixtures of multivariate distributions, of linear regression models, and of factor analyzers, under normal components. We attempt to extend these benefits to the case of skew-normal components. We will show robust methodology based on the joint application of trimming and constraints for different mixture models settings. A drawback of this kind of approaches is related with choosing the input parameters values that this modelling approach requires. We developed different tools now available for assisting the users in getting these values