Inférence statistique dans les modèles bilinéaires périodiques

Abstract

The aim or goal of this work is to study some problems of nonlinear time series. The particular class of models nonlinear that was extensity discussed in the literature is he class of periodic bilinear models. The sufficient conditions for strict periodic stationary et second-order periodic stationary, the inversibility and the existence of higher-order moments are derived. The main goal of this work is the estimation in periodic bilinear model by Least Squares; from some conditions one gets consistency estimators and asymptotic normality. This method gives of results very satisfactory that are shown in the simulation.