January 29th 2025, 4.30 pm, Freihaus FH Lecture Hall 3 (yellow area, 2nd floor) Wiedner Hauptstraße 8-10, 1040 Wien
Prof. Kostantinos Fokianos, öffnet eine externe URL in einem neuen Fenster, University of Cyprus
Title: Count Network Autoregression

Abstract: We study general nonlinear models for time series networks of integer valued data. The vector of high dimensional responses, measured on the nodes of a known network, is regressed non-linearly on its lagged value and on lagged values of the neighboring nodes by employing a smooth link function. We study stability conditions for such multivariate process and develop quasi maximum likelihood inference when the network dimension is increasing. In addition, we study linearity score tests by treating separately the cases of identifiable and non-identifiable parameters. In the case of identifiability, the test statistic converges to a chi-square distribution. When the parameters are not-identifiable, we develop a supremum-type test whose p-values are approximated adequately by employing a feasible bound and bootstrap methodology.