The central aim of this session is to explore modern statistical techniques designed to uncover interesting and informative structures in complex multivariate data. Whether through dimension reduction, latent variable modelling, or clustering, the focus is on identifying patterns that reveal underlying mechanisms or simplify interpretation. The used approaches cover several aspects common in modern data sets: high dimensionality, non-Gaussianity, matrix structure, spatial dependencies and latent group structures. Theoretical contributions will be illustrated by both simulated and real-world examples.
Organized by:
Joni Virta (Finland)Kurtosis-based projection pursuit for matrix-valued data
Invited Speakers:
- Andreas Alfons (The Netherlands)
Sparse Clusterpath Estimation of Gaussian Graphical Models - Jaakko Pere (Finland)
On Stationary Subspace Analysis for Spatio-Temporal Data