Non-Asymptotic Guarantees for Robust Identification of Granger Causality via the LASSO

Abstract

Granger causality is among the widely used data-driven approaches for causal analysis of time series data with applications in various areas including economics, molecular biology, and neuroscience. Two of the main challenges of this methodology are: 1) over-fitting as a result of limited data duration, and 2) correlated process noise as a confounding factor, both leading to errors in identifying the causal influences. Sparse estimation via the LASSO has successfully addressed these challenges for parameter estimation. However, the classical statistical tests for Granger causality resort to asymptotic analysis of ordinary least squares, which require long data durations to be useful and are not immune to confounding effects. In this work, we close this gap by introducing a LASSO-based statistic and studying its non-asymptotic properties under the assumption that the true models admit sparse autoregressive representations. We establish that the sufficient conditions of LASSO also suffice for robust identification of Granger causal influences. We also characterize the false positive error probability of a simple thresholding rule for identifying Granger causal effects. We present simulation studies and application to real data to compare the performance of the ordinary least squares and LASSO in detecting Granger causal influences, which corroborate our theoretical results.

NLGC: Network Localized Granger Causality with Application to MEG Directional Functional Connectivity Analysis

“…In this way, the chosen λ gives a stable solution across the folds. Moreover, once the optimal regularization parameter λ is chosen for the full model, we use the same regularization parameter for all the subsequent reduced models (Das and Babadi, 2021). This way, the cross-validation only needs to be carried out for the full model.…”