Reliable causal discovery in time series

Kurzfassung

In this talk I will, first, introduce the so-called causal discovery task, that is, the task of learning cause-and-effect relationships from observational data. I will then present three algorithms for causal discovery that have been (co-)developed by the Causal Inference research group at the DLR-Institute of Data Science in Jena. These algorithms are specifically designed to deal with the specifics of time series data and are available as part of the Python package tigramite.