Non-parametric segmentation of non-stationary time series

摘要

The non-stationary evolution of observable quantities in complex systems canfrequently be described as a juxtaposition of quasi-stationary spells. Giventhat standard theoretical and data analysis approaches usually rely on theassumption of stationarity, it is important to detect in real time seriesintervals holding that property. With that aim, we introduce a segmentationalgorithm based on a fully non-parametric approach. We illustrate itsapplicability through the analysis of real time series presenting diversedegrees of non-stationarity, thus showing that this segmentation proceduregeneralizes and allows to uncover features unresolved by previous proposalsbased on the discrepancy of low order statistical moments only.