PCMCI theory

PCMCI (Runge et al., 2019) recovers a time-series CPDAG over a set of stationary processes by combining a per-target lagged-parent selection step (PC₁) with a momentary conditional-independence test (MCI) that controls for autocorrelation.

Note

This page is currently a stub. The full treatment, parallel to PC theory, will land in v0.x.x.

References

• Runge, J., Nowack, P., Kretschmer, M., Flaxman, S., & Sejdinovic, D. (2019). Detecting and quantifying causal associations in large nonlinear time series datasets. Science Advances, 5(11), eaau4996.