Doubly censored power-normal regression models with inflation

• Guillermo Martínez-Flórez ^[1] ; Heleno Bolfarine ^[2] ; Héctor W. Gómez ^[3]
  1. [1] Universidad de Córdoba (Colombia)

     Universidad de Córdoba (Colombia)

     Colombia

     
  2. [2] Universidade de São Paulo

     Universidade de São Paulo

     Brasil

     
  3. [3] Universidad de Antofagasta

     Universidad de Antofagasta

     Antofagasta, Chile

     

  Mostrar afiliaciones +
• Localización: Test: An Official Journal of the Spanish Society of Statistics and Operations Research, ISSN-e 1863-8260, ISSN 1133-0686, Vol. 24, Nº. 2, 2015, págs. 265-286
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • In this paper we consider power distributions for modeling proportions or rates with zero and/or one inflation as an alternative to beta regression. The considered model is a mixture between a Bernoulli type process for explaining the zero and/or one responses and a limited power-normal model for explaining the continuous responses. We consider extensions by incorporating covariates in both stages of the modeling process. An advantage of the power-normal distribution is its simple distribution function, easy to deal with in censored situations. The likelihood approach is considered for parameter estimation and model comparison. Given the flexibility of the power-normal distribution, we are able to show, in practical scenarios, the better performance of our proposal over the beta regression model with zero and/or one excess (BIZU).