Title: Stationary subspace analysis: A statistical perspective
Authors: Lea Flumian - Vienna University of Technology (Austria)
Markus Matilainen - University of Turku/Turku PET Centre (Finland)
Klaus Nordhausen - Vienna University of Technology (Austria) [presenting]
Abstract: Multivariate times series occur in many application areas and are challenging to model. A common approach is therefore to assume that the observed time series can be decomposed into latent components with different exploitable properties. In some of these models especially nonstationary components are of interest, and thus the nonstationary subspace should be separated from the stationary subspace which is often referred to as stationary subspace analysis (SSA). Different methods are considered for this purpose and a test suggested to make inference about the dimensions of the subspaces.