Title: On robust change-point detection in time series
Authors: Roland Fried - TU Dortmund University (Germany) [presenting]
Abstract: Some robust techniques are discussed for testing whether a time series is second order stationary. This assumption is commonly applied in time series modelling. So far mainly CUSUM-type tests are applied to test this hypothesis against the alternative of a structural break. We focus on robust tests for testing whether a change of location has occurred and give some ideas how to test robustly for a change of variance or a change of autocorrelation.