The aim of this work is to propose a new multivariate regression model for compositional data, i.e., vectors of proportions. It is based on a mixture of Dirichletdistributed components and it enables many relevant properties for compositional data as well as accounting for positive correlations. Despite the complexity of the model, its special mixture structure provides a greater flexibility and a richer parameterization than the standard Dirichlet regression (DirReg) model and, moreover, guarantees its identifiability. We illustrate the performance and the goodness of fit of our new model through an application to the last Italian elections data.

Di Brisco, A., Ascari, R., Migliorati, S., Ongaro, A. (2019). A new regression model for bounded multivariate responses. In Book of Short Papers SIS2019 (pp.817-822).

A new regression model for bounded multivariate responses

Di Brisco, AM;Ascari, R;Migliorati, S;Ongaro, A
2019

Abstract

The aim of this work is to propose a new multivariate regression model for compositional data, i.e., vectors of proportions. It is based on a mixture of Dirichletdistributed components and it enables many relevant properties for compositional data as well as accounting for positive correlations. Despite the complexity of the model, its special mixture structure provides a greater flexibility and a richer parameterization than the standard Dirichlet regression (DirReg) model and, moreover, guarantees its identifiability. We illustrate the performance and the goodness of fit of our new model through an application to the last Italian elections data.
slide + paper
simplex, mixture model, dirichlet distribution, bayesian inference
English
Smart Statistics for Smart Applications - SIS 2019
2019
Arbia, G; Peluso, S; Pini, A; Rivellini, G
Book of Short Papers SIS2019
9788891915108
2019
817
822
none
Di Brisco, A., Ascari, R., Migliorati, S., Ongaro, A. (2019). A new regression model for bounded multivariate responses. In Book of Short Papers SIS2019 (pp.817-822).
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/10281/233977
Citazioni
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
Social impact