Ganna Fagerberg, dept. of Statistics, Stockholm University
SEMINAR
Date: Wednesday 15 April 2026
Time: 13:00 - 14:00
Location: Campus Albano, lecture room 27, house 4, level 2
Time-Varying Multi-Seasonal ARMA Models with Global-Local Shrinkage Priors
Abstract
Modern time series data are often collected at high frequency and exhibit multiple, evolving seasonal patterns, challenging classical stationarity assumptions. This talk presents a Bayesian framework for time-varying multi-seasonal ARMA models that jointly capture seasonal and non-seasonal dynamics. The approach uses global–local shrinkage priors to adaptively control time variation. Efficient state-space inference and scalable computation are discussed, with applications demonstrating improved forecasting performance and interpretability.
Last updated: 2026-04-13
Source: Department of Statistics