Abstract
Inferring causal relations from observational time series data is a key problem across science and engineering whenever experimental interventions are infeasible or unethical. Increasing data availability over the past few decades has spurred the development of a plethora of causal discovery methods, each addressing particular challenges of this difficult task. In this paper, we focus on an important challenge that is at the core of time series causal discovery: regime-dependent causal relations. Often dynamical systems feature transitions depending on some, often persistent, unobserved background regime, and different regimes may exhibit different causal relations. Here, we assume a persistent and discrete regime variable leading to a finite number of regimes within which we may assume stationary causal relations. To detect regime-dependent causal relations, we combine the conditional independence-based PCMCI method [based on a condition-selection step (PC) followed by the momentary conditional independence (MCI) test] with a regime learning optimization approach. PCMCI allows for causal discovery from high-dimensional and highly correlated time series. Our method, Regime-PCMCI, is evaluated on a number of numerical experiments demonstrating that it can distinguish regimes with different causal directions, time lags, and sign of causal links, as well as changes in the variables' autocorrelation. Furthermore, Regime-PCMCI is employed to observations of El Niño Southern Oscillation and Indian rainfall, demonstrating skill also in real-world datasets.
Highlights
Understanding causal relationships1,2 among different processes is a ubiquitous task in many scientific disciplines as well as engineering
The ground-truth regime evolution and networks are shown in the top part of panels a and b in Fig. 2; in the middle part of both panels, their Regime-PCMCI reconstruction is shown; and in the bottom part, the difference between reconstructed and true regimes is presented to visually inspect the accuracy
Causal discovery is emerging as an important framework across many disciplines in science and engineering, but each discipline has particular challenges that novel methods need to address
Summary
Understanding causal relationships among different processes is a ubiquitous task in many scientific disciplines as well as engineering (e.g., in the context of climate research, econometrics, molecular, and animal group dynamics). All that is often given is a set of time series describing these processes with no specific knowledge about the direction and form scitation.org/journal/cha of their causal relationships available. The challenge, termed causal discovery, is to reconstruct the underlying graph of causal relationships from time series data.. The processes that generated the data can be modeled in the framework of structural causal models (SCMs) to further understand causal relations, predict the effect of interventions, and for forecasting. Today’s ever-growing abundance of time series datasets promises many application scenarios for the numerous datadriven causal discovery methods. Causal knowledge cannot be gained from data alone and each method comes with its particular set of assumptions about properties of the underlying processes and the observed data. Runge et al. recently provided an overview of current methodological frameworks, their application scenarios, and open challenges
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
