Colloquium: Approximation of the local Bayes empirical factor for the comparison of DLM ARMA(1,1) and DLM ARMA(1,0) models by introducing a perturbed evolution matrix G

In this talk, we present an alternative to obtaining an approximation of the Local Bayes Empirical Factor $latex B_{j,i}$ (based on real training data) for the comparison
of the Nested Dynamic Linear Models (DLM’s) $latex M_j$ = DLM ARMR(1,1), and $latex M_i$ = DLM ARMA(1,0).