Partial Distance Correlation Screening for High Dimensional Time Series

TL;DR

This paper develops model-free screening methods based on partial distance correlation tailored for high-dimensional time series data, addressing the unique dependence structures in such datasets.

Contribution

It introduces novel variable screening techniques for time series that leverage partial distance correlation and proves their sure screening properties.

Findings

Methods perform well in simulations.

Applicable to univariate and multivariate models.

Effective in macroeconomic forecasting.

Abstract

High dimensional time series datasets are becoming increasingly common in various fields such as economics, finance, meteorology, and neuroscience. Given this ubiquity of time series data, it is surprising that very few works on variable screening discuss the time series setting, and even fewer works have developed methods which utilize the unique features of time series data. This paper introduces several model free screening methods based on the partial distance correlation and developed specifically to deal with time dependent data. Methods are developed both for univariate models, such as nonlinear autoregressive models with exogenous predictors (NARX), and multivariate models such as linear or nonlinear VAR models. Sure screening properties are proved for our methods, which depend on the moment conditions, and the strength of dependence in the response and covariate processes, amongst other factors. Dependence is quantified by functional dependence measures (Wu [Proc. Natl. Acad. Sci. USA 102 (2005) 14150-14154]) and $\beta$-mixing coefficients, and the results rely on the use of Nagaev and Rosenthal type inequalities for dependent random variables. Finite sample performance of our methods is shown through extensive simulation studies, and we include an application to macroeconomic forecasting.