I am the king of the world! But am I?
So far, we’ve seen what we can achieve by using causal models on observational data and we promised that we’ll see them do even more impressive things. All this might feel pretty powerful!
In this section, we’re going to discuss important challenges that we might face when using causal inference methods in practice.
We’ll start by sketching a broader context. After that, we’ll explicitly define the concept of identifiability. Finally, we’ll discuss some popular challenges faced by practitioners:
- A lack of a priori knowledge of causal graphs
- Insufficient sample sizes
- Difficulties with verifying the assumptions
In between
Ancient Greek mythology provides us with a story of Ikaros (also known as Icarus) and his father, the Athenian inventor Daidalos (also known as Daedalus). Daidalos wants to escape from Crete – a Greek island where he’s trapped. He builds two...