Constraint-based causal discovery in epidemiology

I will discuss the challenges in applying constraint-based causal discovery to typical empidemiological data. This includes dealing with mixed measurement scales, repeated measures, missing values, conflict resolutions, bias in the bootstrap, and time structure. A focus will be on the latter aspect: the efficiency gains of a temporal PC-algorithm and how insights can be gained from characterising the equivalence class under tiered background knowledge.




Previous Speaker Next Speaker

  • Keine Stichwörter