Modelos lineares generalizados simÃtricos heterocedÃsticos

AUTOR(ES)

Maria LÃdia Coco Terra

DATA DE PUBLICAÇÃO

2009

RESUMO

The heteroscedastic symmetrical generalized linear models class is describe for a response that belongs to a symmetrical class of distributions, a linear predictor relate to a location parameter and a linear predictor associate to a dispersion parameter, both link functions. These models are natural extensions of symmetrical generalized linear models propose by Lobos (2004), where only a location parameter is associate to a one-to-one continuously differentiable positive link function. First of all, in this work we present some results for heteroscedastic symmetrical generalized linear models. Then, asymptotic tests to evaluate homogeneity are proposed as well as diagnostic methods based on local influence. Some examples with real data are discuss to illustrate the developed theory.

ASSUNTO(S)

modelos lineares generalizados simÃtricos heteroscedÃsticos influÃncia local influÃncia local modelos lineares generalizados simÃtricos heteroscedÃsticos parÃmetro de escala parÃmetro de escala estatistica

Documentos Relacionados

• CorreÃÃo de Bartlett nos modelos nÃo-lineares simÃtricos heteroscedÃsticos
• AperfeiÃoamento de testes de hipÃteses para modelos nÃo-lineares simÃtricos
• DiagnÃstico em modelos simÃtricos de regressÃo
• Bayesian multiple comparisons in homocedastic and heterocedastic normal models.
• Modelos lineares generalizados aplicados à filariose bancroftiana.