Online Program Home
My Program

Abstract Details

Activity Number: 460 - Advances in Time Series Methodology
Type: Topic Contributed
Date/Time: Wednesday, July 31, 2019 : 8:30 AM to 10:20 AM
Sponsor: Business and Economic Statistics Section
Abstract #305338
Title: Post Selection Inference for High-Dimensional Time Series
Author(s): Anand Vidyashankar* and Jeffrey Collamore
Companies: George Mason University and University of Copenhagen
Keywords: ARMA(p,q); return times; sparse model; regerative process; least squares; post-selection estimators

High-dimensional time series arise in several contemporary applications in finance and engineering. In ARMA type models, assuming sparsity, it is common to use regularization based methods such as LASSO and SCAD to perform model/variable selection. However, the effect of performing model selection on the resulting inference is largely unknown. In this presentation, we focus on stationary ARMA-type models and describe a new regeneration based inferential procedure that accounts for model selection uncertainty. We establish the consistency and asymptotic normality of the proposed post-selection estimators. Simulation results and data analyses are included to demonstrate the proposed methods. We bring to fore the importance of the return times over regeneration times of the Markov chain associated with these processes in the current setting.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program