GAMLSS for high-variability data: an application to liver fibrosis case.

Research output: Contribution to journalArticlepeer-review


In this paper, we propose management of theproblem caused by overdispersed data by applying the generalized additive modelfor location, scale and shape framework (GAMLSS) as introduced by Rigby andStasinopoulos (2005). The idea of using a GAMLSS approach for handling ourproblem comes from the idea of Aitkin (1996) consisting in the use of an EM maximumlikelihood estimation algorithm (Dempster, Laird, and Rubin, 1977) to dealwith overdispersed generalized linear models (GLM). As in the GLM case, the algorithmis initially derived as a form of Gaussian quadrature assuming a normalmixing distribution. The GAMLSS specification allows the extension of the Aitkinalgorithm to probability distributions not belonging to the exponential family. Inparticular, aim of this work is to show the importance of using a GAMLSS strutcurewhen a mixture is used to provide a natural representation of heterogeneity in a finitenumber of latent classes (Celeux and Diebolt, 1992).
Original languageEnglish
Number of pages31
Publication statusPublished - 2020

All Science Journal Classification (ASJC) codes

  • Statistics and Probability
  • Statistics, Probability and Uncertainty


Dive into the research topics of 'GAMLSS for high-variability data: an application to liver fibrosis case.'. Together they form a unique fingerprint.

Cite this