High-recall causal discovery for autocorrelated time series with latent confounders

TL;DR

This paper introduces a novel causal discovery method tailored for autocorrelated time series with latent confounders, significantly improving recall over existing techniques by leveraging an iterative approach and new orientation rules.

Contribution

The paper proposes an order-independent, sound, and complete iterative causal discovery method that enhances effect size and recall in autocorrelated time series with latent confounders.

Findings

Achieves higher recall than existing methods in simulations

Performance improves with stronger autocorrelation

Maintains false positive rate at desired level

Abstract

We present a new method for linear and nonlinear, lagged and contemporaneous constraint-based causal discovery from observational time series in the presence of latent confounders. We show that existing causal discovery methods such as FCI and variants suffer from low recall in the autocorrelated time series case and identify low effect size of conditional independence tests as the main reason. Information-theoretical arguments show that effect size can often be increased if causal parents are included in the conditioning sets. To identify parents early on, we suggest an iterative procedure that utilizes novel orientation rules to determine ancestral relationships already during the edge removal phase. We prove that the method is order-independent, and sound and complete in the oracle case. Extensive simulation studies for different numbers of variables, time lags, sample sizes, and further cases demonstrate that our method indeed achieves much higher recall than existing methods for the case of autocorrelated continuous variables while keeping false positives at the desired level. This performance gain grows with stronger autocorrelation. At https://github.com/jakobrunge/tigramite we provide Python code for all methods involved in the simulation studies.