Activity Number:
|
509
- Recent Advances in High-Dimensional Time Series Analysis
|
Type:
|
Topic Contributed
|
Date/Time:
|
Thursday, August 11, 2022 : 8:30 AM to 10:20 AM
|
Sponsor:
|
Business and Economic Statistics Section
|
Abstract #323277
|
|
Title:
|
Inference for Location of Change Points in High-Dimensional Non-Stationary Vector Auto-Regressive Models
|
Author(s):
|
Abolfazl Safikhani* and Abhishek Kaul and Yue Bai
|
Companies:
|
University of Florida and Washington State University and University of Florida
|
Keywords:
|
inference;
change point;
time series;
high-dimension;
VAR
|
Abstract:
|
Piece-wise stationary Vector Auto-Regressive models (VAR) are among well-known and useful models in time series analysis. Existing methods provide sub-optimal estimators to detect location of change/break points in high-dimensional VAR models due to existence of terms such as total sparsity of transition matrices and logarithm of the number of time series components in the consistency rate. In this talk, we study a refitted least squares estimator for change point parameters in high-dimensional VAR models with sparse model parameters. We show that the newly defined estimator reaches optimal rate of convergence and the corresponding rate for relative location of change points reaches O(1/T) for certain non-vanishing jump sizes, where T is the sample size. Further, the limiting distribution of the proposed estimate is obtained under both vanishing and non-vanishing jump sizes, thereby allowing construction of confidence intervals for change point parameters. The proposed methodology is tested empirically over different synthetic data sets while an application to analyzing an EEG data set is also provided.
|
Authors who are presenting talks have a * after their name.