Home > Research > Publications & Outputs > Online non-parametric changepoint detection wit...

Electronic data

  • NUNC_CSDA_FINAL

    Rights statement: This is the author’s version of a work that was accepted for publication in Computational Statistics & Data Analysis. Changes resulting from the publishing process, such as peer review, editing, corrections, structural formatting, and other quality control mechanisms may not be reflected in this document. Changes may have been made to this work since it was submitted for publication. A definitive version was subsequently published in Computational Statistics & Data Analysis, 177, 2022 DOI: 10.1016/j.csda.2022.107551

    Accepted author manuscript, 624 KB, PDF document

    Embargo ends: 3/08/23

    Available under license: CC BY-NC-ND: Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License

Links

Text available via DOI:

View graph of relations

Online non-parametric changepoint detection with application to monitoring operational performance of network devices

Research output: Contribution to Journal/MagazineJournal articlepeer-review

E-pub ahead of print
Article number107551
<mark>Journal publication date</mark>31/01/2023
<mark>Journal</mark>Computational Statistics and Data Analysis
Volume177
Number of pages13
Publication StatusE-pub ahead of print
Early online date3/08/22
<mark>Original language</mark>English

Abstract

Motivated by a telecommunications application where there are few computational constraints, a novel nonparametric algorithm, NUNC, is introduced to perform an online detection for changes in the distribution of data. Two variants are considered: the first, NUNC Local, detects changes within a sliding window. Conversely, NUNC Global, compares the current window of data to all of the historic information seen so far and makes use of an efficient update step so that this historic information does not need to be stored. To explore the properties of both algorithms, both real and simulated datasets are analysed. Furthermore, a theoretical result for the choice of test threshold to control the false alarm rate is presented, a result that could be applied in other binary segmentation change detection settings.

Bibliographic note

This is the author’s version of a work that was accepted for publication in Computational Statistics & Data Analysis. Changes resulting from the publishing process, such as peer review, editing, corrections, structural formatting, and other quality control mechanisms may not be reflected in this document. Changes may have been made to this work since it was submitted for publication. A definitive version was subsequently published in Computational Statistics & Data Analysis, 177, 2022 DOI: 10.1016/j.csda.2022.107551