Transformações em modelos de séries temporais / Transformations in time series models

AUTOR(ES)
FONTE

IBICT - Instituto Brasileiro de Informação em Ciência e Tecnologia

DATA DE PUBLICAÇÃO

21/05/2012

RESUMO

Cordeiro and Andrade (2009) incorporate the idea of transforming the response variable to the GARMA model, generalized autoregressive moving average, introduced by Benjamin et al. (2003), thus developing the TGARMA model, transformed generalized autoregressive moving average. The goal of this thesis is to develop the TGARMA model introduced by Cordeiro and Andrade (2009) for symmetric continuous conditional distributions and a possible non-linear structure for the mean that enables the fitting of a wide range of models to several data types. When the assumption of homoscedasticity is not verified, heteroscedastic models are proposed. Throughout this thesis, we derive an iterative process for fitting the parameters of the models by maximum likelihood. We produce a simple formula to estimate the parameter which defines the transformation of the response variable and the moments of the original dependent variable which generalize previous published results. For the homoscedastic model, we discuss inference, we propose a diagnostic analysis and define a standardized residuals. Finally, to illustrate the theory developed, we use real data sets and we evaluate the results developed through simulations studies.

ASSUNTO(S)

distribuição simétrica parâmetro de transformação séries temporais symmetric distribution transformation parameter time series

Documentos Relacionados