Activity Number:
|
185
|
Type:
|
Contributed
|
Date/Time:
|
Monday, August 1, 2016 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Section on Nonparametric Statistics
|
Abstract #319207
|
View Presentation
|
Title:
|
Nonparametric Change-Point Detection in Multivariate Nonstationary Time Series
|
Author(s):
|
Raanju Ragavendar Sundararajan* and Mohsen Pourahmadi
|
Companies:
|
Texas A&M University and Texas A&M University
|
Keywords:
|
multivariate time series ;
spectral matrix ;
spectral decomposition ;
strong motion data
|
Abstract:
|
Detecting change points in multivariate time series is an important problem with numerous applications. Much of change point literature is on tackling this problem in the univariate case or is parametric in nature. We develop a nonparametric method to detect multiple change points in multivariate piecewise stationary processes when the locations and number of change points are unknown. Based on a test statistic that measures differences in the spectral density matrices through the L_2 norm, we propose a two stage procedure. The first stage identifies potential change points using a sequential test and the second stage tests for the significance of each of the potential change points. The asymptotic properties of the test for significant change points under the null and alternative hypothesis are derived. Monte Carlo simulation of values of a stationary process given its spectral density matrix is used to obtain critical values of the test statistic under the null. We illustrate the better performance of our method in comparison to some of the recent methods through a few simulation examples and discuss and an application of our method in seismology.
|
Authors who are presenting talks have a * after their name.