We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

math.ST

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Mathematics > Statistics Theory

Title: A Note on Online Change Point Detection

Abstract: We investigate sequential change point estimation and detection in univariate nonparametric settings, where a stream of independent observations from sub-Gaussian distributions with a common variance factor and piecewise-constant but otherwise unknown means are collected. We develop a simple CUSUM-based methodology that provably control the probability of false alarms or the average run length while minimizing, in a minimax sense, the detection delay. We allow for all the model parameters to vary in order to capture a broad range of levels of statistical hardness for the problem at hand. We further show how our methodology is applicable to the case in which multiple change points are to be estimated sequentially.
Subjects: Statistics Theory (math.ST)
Cite as: arXiv:2006.03283 [math.ST]
  (or arXiv:2006.03283v3 [math.ST] for this version)

Submission history

From: Yi Yu [view email]
[v1] Fri, 5 Jun 2020 08:06:26 GMT (100kb)
[v2] Thu, 29 Oct 2020 18:25:25 GMT (34kb)
[v3] Fri, 13 Nov 2020 10:42:40 GMT (34kb)

Link back to: arXiv, form interface, contact.