Methods of causal inference from observational time series data are recently more and more being adopted in geoscientific research. Modern approaches coming from statistics and machine learning provide powerful tools to extract causal knowledge given only multivariate time series. This course will introduce basic concepts of causal inference and explain in more detail a recently developed approach based on graphical models and causal discovery algorithms [1,2].

Methods of causal inference from observational time series data are recently more and more being adopted in geoscientific research. Modern approaches coming from statistics and machine learning provide powerful tools to extract causal knowledge given only multivariate time series. This course will introduce basic concepts of causal inference and explain in more detail a recently developed approach based on graphical models and causal discovery algorithms [1,2].

The course is aimed at geoscientists who wish to learn more about the advantages and limitations of such approaches. Participants can get hands-on experience with the open-source Python package "Tigramite" and bring their own data.

Code:
https://github.com/jakobrunge/tigramite

References:
[1] Runge, J., Petoukhov, V., & Kurths, J. (2014). Quantifying the Strength and Delay of Climatic Interactions: The Ambiguities of Cross Correlation and a Novel Measure Based on Graphical Models. Journal of Climate, 27(2), 720–739. doi:10.1175/JCLI-D-13-00159.1 [2] Runge, J., Heitzig, J., Petoukhov, V., & Kurths, J. (2012). Escaping the Curse of Dimensionality in Estimating Multivariate Transfer Entropy. Physical Review Letters, 108(25), 258701. doi:10.1103/PhysRevLett.108.258701