Title: Change-point detection based on weighted two-sample U-statistics
Authors: Kata Vuk - Ruhr-University Bochum (Germany) [presenting]
Herold Dehling - Ruhr-University Bochum (Germany)
Martin Wendler - University of Greifswald (Germany)
Abstract: A robust change-point test is considered which is based on weighted two-sample U-statistics. We focus on short range dependent data, more precisely on data that can be represented as functionals of a mixing process. In this way, most examples from time series analysis are covered. Under the hypothesis that no change occurs, the limit distribution of the considered test statistic is derived. Under the alternative of a change-point with constant height, we derive consistency. The considered test is sensitive on tails, which means that early and late changes can be better detected. To illustrate the results and to investigate the power of the test, we will give some simulation results.