Change-points detection for variance piecewise constant models

Research output: Contribution to journalArticlepeer-review

5 Citations (Scopus)

Abstract

A new approach based on the fit of a generalized linear regression model is introduced for detecting change-points in the variance of heteroscedastic Gaussian variables, with piecewise constant variance function. This approach overcome some limitations of both exact and approximate well-known methods that are based on successive application of search and tend to overestimate the real number of changes in the variance of the series. The proposed method just requires the computation of a gamma GLM with log-link, resulting in a very efficient algorithm even with large sample size and many change points to be estimated.
Original languageEnglish
Pages (from-to)437-448
Number of pages12
JournalCOMMUNICATIONS IN STATISTICS. SIMULATION AND COMPUTATION
Volume41
Publication statusPublished - 2012

All Science Journal Classification (ASJC) codes

  • Statistics and Probability
  • Modelling and Simulation

Fingerprint

Dive into the research topics of 'Change-points detection for variance piecewise constant models'. Together they form a unique fingerprint.

Cite this