Shifted-Dirichlet Regression vs Simplicial Regression:: a comparison
-
Autores: G.S. Monti, Glòria Mateu Figueras
, Vera Pawlowsky Glahn , Juan José Egozcue Rubí
- Localización: Proceedings of the 6th International Workshop on Compositional Data Analysis: Girona, 1-7 de juny de 2015 / coord. por Santiago Thió Fernández de Henestrosa , Josep Antoni Martín Fernández, 2015, ISBN 978-84-8458-451-3
- Idioma: inglés
- Enlaces
-
Resumen
In Monti et al. (2014) a compositional model based on Scaled–Dirichlet regression was studied, following the approach suggested by Campbell and Mosimann (1987), and also studied by Hijazi and Jernigan (2009). The Scaled–Dirichlet distribution (Monti et al., 2011) is one of the generalizations of the Dirichlet distribution. Using an approach based on the Aitchison geometry and a suitable measure on the unit simplex S D, Monti et al. (2011) show that the Scaled–Dirichlet distribution can be viewed as the distribution of a perturbed random composition with Dirichlet density. The Scaled–Dirichlet regression model is obtained by allowing its parameters to change linearly with covariates, in order to assess the effects of covariates on the relative contributions of different components in a composition.
