We propose a class of semi-constrained models for clustering ordinal and continuous data. Ordinal variables are assumed to be a discretization of some latent continuous variables jointly distribuited with the observed continuous variables as a finite mixture of Gaussians. Parsimonious modeling is obtained by reparameterizing the covariance matrices in terms of factor analysis models semi-constrained across the components. Parameter estimation is carried out using a EM-type algorithm to maximize a composite log-likelihood. The proposal is evaluated through a simulation study and an application to real data.
Semi-constrained model-based clustering of mixed-type data using a composite likelihood approach / Rocci, Roberto; Ranalli, Monia. - (2021), pp. 408-411. (Intervento presentato al convegno Cladag 2021 tenutosi a Firenze).
Semi-constrained model-based clustering of mixed-type data using a composite likelihood approach
Rocci Roberto;Ranalli Monia
2021
Abstract
We propose a class of semi-constrained models for clustering ordinal and continuous data. Ordinal variables are assumed to be a discretization of some latent continuous variables jointly distribuited with the observed continuous variables as a finite mixture of Gaussians. Parsimonious modeling is obtained by reparameterizing the covariance matrices in terms of factor analysis models semi-constrained across the components. Parameter estimation is carried out using a EM-type algorithm to maximize a composite log-likelihood. The proposal is evaluated through a simulation study and an application to real data.File | Dimensione | Formato | |
---|---|---|---|
Rocci_semi-constrained_2021.pdf
solo gestori archivio
Tipologia:
Versione editoriale (versione pubblicata con il layout dell'editore)
Licenza:
Tutti i diritti riservati (All rights reserved)
Dimensione
725.74 kB
Formato
Adobe PDF
|
725.74 kB | Adobe PDF | Contatta l'autore |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.