Online Program Home
  My Program

Abstract Details

Activity Number: 98 - Statistical Learning for Dependent Data
Type: Invited
Date/Time: Monday, July 31, 2017 : 8:30 AM to 10:20 AM
Sponsor: Business and Economic Statistics Section
Abstract #322331
Title: Estimation of Sparse Vector Autoregressive Moving Averages
Author(s): David S Matteson* and Ines Wilms and Jacob Bien
Companies: Cornell University and KU Leuven and Cornell University
Keywords: time series ; big data ; sparsity ; forecasting ; multivariate ; econometrics

The Vector AutoRegressive Moving Average (VARMA) model is a fundamental tool for modeling multivariate time series. Recently, a growing interest has arisen in high-dimensional models, where the number of marginal time series is increasingly large. However, as the number of time series increases, the VARMA model becomes heavily overparameterized. For such high-dimensional VARMA models, estimation is generally intractable. In this setting, the high-dimensional Vector AutoRegression (VAR) model has been favored, in both theory and practice. We propose adapting modern regularization methods to estimate high-dimensional VARMA models. Our estimation method is sparse, meaning many model parameters are estimated as exactly zero. The proposed framework has good estimation and forecast accuracy under numerous simulation settings. We illustrate the forecast performance of the sparse VARMA models for several application domains, including macro-economic forecasting, demand forecasting and volatility forecasting. The sparse VARMA estimator gives parsimonious forecast models that lead to important gains in relative forecast accuracy.

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

Back to the full JSM 2017 program

Copyright © American Statistical Association