Validity of Time Reversal for Testing Granger Causality

Irene Winkler, Danny Panknin, Daniel Bartz, Klaus Muller, Stefan Haufe

Research output: Contribution to journalArticle

18 Citations (Scopus)

Abstract

Inferring causal interactions from observed data is a challenging problem, especially in the presence of measurement noise. To alleviate the problem of spurious causality, Haufe (2013) proposed to contrast measures of information flow obtained on the original data against the same measures obtained on time-reversed data. They show that this procedure, time-reversed Granger causality (TRGC), robustly rejects causal interpretations on mixtures of independent signals. While promising results have been achieved in simulations, it was so far unknown whether time reversal leads to valid measures of information flow in the presence of true interaction. Here, we prove that, for linear finite-order autoregressive processes with unidirectional information flow between two variables, the application of time reversal for testing Granger causality indeed leads to correct estimates of information flow and its directionality. Using simulations, we further show that TRGC is able to infer correct directionality with similar statistical power as the net Granger causality between two variables, while being much more robust to the presence of measurement noise.

Original languageEnglish
Article number7412766
Pages (from-to)2746-2760
Number of pages15
JournalIEEE Transactions on Signal Processing
Volume64
Issue number11
DOIs
Publication statusPublished - 2016 Jun 1

Fingerprint

Testing

Keywords

  • Granger causality
  • noise
  • time reversal
  • TRGC

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Signal Processing

Cite this

Winkler, I., Panknin, D., Bartz, D., Muller, K., & Haufe, S. (2016). Validity of Time Reversal for Testing Granger Causality. IEEE Transactions on Signal Processing, 64(11), 2746-2760. [7412766]. https://doi.org/10.1109/TSP.2016.2531628

Validity of Time Reversal for Testing Granger Causality. / Winkler, Irene; Panknin, Danny; Bartz, Daniel; Muller, Klaus; Haufe, Stefan.

In: IEEE Transactions on Signal Processing, Vol. 64, No. 11, 7412766, 01.06.2016, p. 2746-2760.

Research output: Contribution to journalArticle

Winkler, I, Panknin, D, Bartz, D, Muller, K & Haufe, S 2016, 'Validity of Time Reversal for Testing Granger Causality', IEEE Transactions on Signal Processing, vol. 64, no. 11, 7412766, pp. 2746-2760. https://doi.org/10.1109/TSP.2016.2531628
Winkler I, Panknin D, Bartz D, Muller K, Haufe S. Validity of Time Reversal for Testing Granger Causality. IEEE Transactions on Signal Processing. 2016 Jun 1;64(11):2746-2760. 7412766. https://doi.org/10.1109/TSP.2016.2531628
Winkler, Irene ; Panknin, Danny ; Bartz, Daniel ; Muller, Klaus ; Haufe, Stefan. / Validity of Time Reversal for Testing Granger Causality. In: IEEE Transactions on Signal Processing. 2016 ; Vol. 64, No. 11. pp. 2746-2760.
@article{cc6b4c0982154e8c806a498f34acba3d,
title = "Validity of Time Reversal for Testing Granger Causality",
abstract = "Inferring causal interactions from observed data is a challenging problem, especially in the presence of measurement noise. To alleviate the problem of spurious causality, Haufe (2013) proposed to contrast measures of information flow obtained on the original data against the same measures obtained on time-reversed data. They show that this procedure, time-reversed Granger causality (TRGC), robustly rejects causal interpretations on mixtures of independent signals. While promising results have been achieved in simulations, it was so far unknown whether time reversal leads to valid measures of information flow in the presence of true interaction. Here, we prove that, for linear finite-order autoregressive processes with unidirectional information flow between two variables, the application of time reversal for testing Granger causality indeed leads to correct estimates of information flow and its directionality. Using simulations, we further show that TRGC is able to infer correct directionality with similar statistical power as the net Granger causality between two variables, while being much more robust to the presence of measurement noise.",
keywords = "Granger causality, noise, time reversal, TRGC",
author = "Irene Winkler and Danny Panknin and Daniel Bartz and Klaus Muller and Stefan Haufe",
year = "2016",
month = "6",
day = "1",
doi = "10.1109/TSP.2016.2531628",
language = "English",
volume = "64",
pages = "2746--2760",
journal = "IEEE Transactions on Signal Processing",
issn = "1053-587X",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "11",

}

TY - JOUR

T1 - Validity of Time Reversal for Testing Granger Causality

AU - Winkler, Irene

AU - Panknin, Danny

AU - Bartz, Daniel

AU - Muller, Klaus

AU - Haufe, Stefan

PY - 2016/6/1

Y1 - 2016/6/1

N2 - Inferring causal interactions from observed data is a challenging problem, especially in the presence of measurement noise. To alleviate the problem of spurious causality, Haufe (2013) proposed to contrast measures of information flow obtained on the original data against the same measures obtained on time-reversed data. They show that this procedure, time-reversed Granger causality (TRGC), robustly rejects causal interpretations on mixtures of independent signals. While promising results have been achieved in simulations, it was so far unknown whether time reversal leads to valid measures of information flow in the presence of true interaction. Here, we prove that, for linear finite-order autoregressive processes with unidirectional information flow between two variables, the application of time reversal for testing Granger causality indeed leads to correct estimates of information flow and its directionality. Using simulations, we further show that TRGC is able to infer correct directionality with similar statistical power as the net Granger causality between two variables, while being much more robust to the presence of measurement noise.

AB - Inferring causal interactions from observed data is a challenging problem, especially in the presence of measurement noise. To alleviate the problem of spurious causality, Haufe (2013) proposed to contrast measures of information flow obtained on the original data against the same measures obtained on time-reversed data. They show that this procedure, time-reversed Granger causality (TRGC), robustly rejects causal interpretations on mixtures of independent signals. While promising results have been achieved in simulations, it was so far unknown whether time reversal leads to valid measures of information flow in the presence of true interaction. Here, we prove that, for linear finite-order autoregressive processes with unidirectional information flow between two variables, the application of time reversal for testing Granger causality indeed leads to correct estimates of information flow and its directionality. Using simulations, we further show that TRGC is able to infer correct directionality with similar statistical power as the net Granger causality between two variables, while being much more robust to the presence of measurement noise.

KW - Granger causality

KW - noise

KW - time reversal

KW - TRGC

UR - http://www.scopus.com/inward/record.url?scp=84968765214&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84968765214&partnerID=8YFLogxK

U2 - 10.1109/TSP.2016.2531628

DO - 10.1109/TSP.2016.2531628

M3 - Article

AN - SCOPUS:84968765214

VL - 64

SP - 2746

EP - 2760

JO - IEEE Transactions on Signal Processing

JF - IEEE Transactions on Signal Processing

SN - 1053-587X

IS - 11

M1 - 7412766

ER -