Title: Speeding up inference for high dimensional time series models
Authors: Daniel Paulin - University of Edinburgh (United Kingdom) [presenting]
Abstract: Techniques for fitting multivariate VARMA models will be overviewed. Despite the extensive work in this area, the computation cost of existing methods becomes very high for multivariate time series with more than ten variables. To address this challenge, we propose a new technique that offers improvements both in computational efficiency and predictive accuracy.