Processo INAR(1) com estrutura sazonal para séries temporais de valores inteiros com sobredispersão

Abstract

The study of integer-valued time series models is increasingly present in the literature. It is common in practice for series to contain a seasonal component, and unlike continuous models, counting processes with a seasonal structure have not received much attention in the literature so far. The aims of this paper are: to introduce a new autoregressive model for non-negative integer-valued time series with seasonal structure which are overdispersed, to define the main process properties, to study the estimation methods of the parameters of the proposed model, these methods are Yule-Walker, conditional least squares and conditional maximum likelihood estimators, comparing them in a simulation study and finally applying the proposed model to a real data set. We compared the new model with models already proposed in the literature, and, the model proposed in this paper presented a better fit.