Title: A linear mixed-effect state space model to discriminate unobservable structural components
Authors: Marco Costa - University of Aveiro (Portugal) [presenting]
Magda Monteiro - University of Aveiro (Portugal)
Abstract: A linear mixed-effect state space model is proposed in order to discriminate monthly environmental time series through the prediction of unobservable components (for instance, trend and seasonality components). This model incorporates both fixed and stochastic effects and it allows the application of the Kalman filter and the Kalman smoother predictors. The parameters estimation is discussed and it is performed with both Gaussian maximum likelihood method and distribution-free estimators in a two-step procedure. The application of the Kalman smoother algorithm allows obtaining predictions of the structural components. The proposed approach is illustrated in the discrimination of the water monitoring sites using the monthly dissolved oxygen concentration dataset between January 2002 and May 2013, in the hydrological basin of the river Vouga, in Portugal. The water monitoring sites are discriminated through the structural components by a hierarchical agglomerative clustering procedure.