The talk will discuss the use of finite mixtures of multivariate skew- normal distributions as an approach to model-based clustering. The main motivation for using skew-normal densities as the components of the mixture is that they provide a formulation for parsimonius modelling of data clusters having convex but not elliptical shape. An EM algorithm for maximum likelihood estimation will be described, and some examples of application on both real and simulated data sets will be shown.
Some results on mixtures of skew-normal distributions / A. Capitanio; C. Viroli. - STAMPA. - (2009), pp. 229-230. (Intervento presentato al convegno 6th St. Petersburg Workshop on Simulation tenutosi a St. Petersburg nel June 28 - July 4).
Some results on mixtures of skew-normal distributions
CAPITANIO, ANTONELLA;VIROLI, CINZIA
2009
Abstract
The talk will discuss the use of finite mixtures of multivariate skew- normal distributions as an approach to model-based clustering. The main motivation for using skew-normal densities as the components of the mixture is that they provide a formulation for parsimonius modelling of data clusters having convex but not elliptical shape. An EM algorithm for maximum likelihood estimation will be described, and some examples of application on both real and simulated data sets will be shown.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
