Schedule 25/26

High-Dimensional Statistical Learning for (Financial and Economic) Time Series: A Bayesian's View

Gregor Kastner, University of Klagenfurt

Abstract: We address typical challenges that arise when dealing with multivariate and potentially high-dimensional time series data. By taking a Bayesian stance and carefully balancing model complexity with parameter sparsity, remedies to (some of) those challenges are proposed. A particular focus is placed on time-varying (co)variance estimation and prediction. In addition, we shed light on the computational issues that appear when implementing our proposals. Throughout the talk, the efficacy of the models and methods proposed is illustrated through applications in economics, finance, and beyond. 

TBA

Alain Hecq, Maastricht University