robcp: Robust Change-Point Tests

Provides robust methods to detect change-points in uni- or multivariate time series. They can cope with corrupted data and heavy tails. Focus is on the detection of abrupt changes in location, but changes scale or dependence structure can be detected as well. This package provides tests for change detection in uni- and multivariate time series based on Huberized versions of CUSUM tests proposed in Duerre and Fried (2019) <arXiv:1905.06201>, and tests for change detection in univariate time series based on 2-sample U-statistics or 2-sample U-quantiles as proposed by Dehling et al. (2015) <DOI:10.1007/978-1-4939-3076-0_12> and Dehling, Fried and Wendler (2020) <DOI:10.1093/biomet/asaa004>. Furthermore, the packages provides tests on changes in the scale or the correlation as proposed in Gerstenberger, Vogel and Wendler (2020) <DOI:10.1080/01621459.2019.1629938>, Dehling et al. (2017) <DOI:10.1017/S026646661600044X>, and Wied et al. (2014) <DOI:10.1016/j.csda.2013.03.005>.

Getting started

Package details

AuthorSheila Goerz [aut, cre], Alexander Duerre [aut]
MaintainerSheila Goerz <>
Package repositoryView on CRAN
Installation Install the latest version of this package by entering the following in R:

Try the robcp package in your browser

Any scripts or data that you put into this service are public.

robcp documentation built on Sept. 16, 2022, 5:05 p.m.