Interventions versus counterfactuals

[drbenvincent]

Notebook proposal

Title: Interventions versus counterfactuals

Why should this notebook be added to pymc-examples?

There is a lot of excitement about causal inference at the moment. Interventions and Counterfactuals map on to Pearl's Rung 2 and Rung 3 of the causal ladder. So having a notebook that succinctly explains the concepts and how to implement these in PyMC would be valuable.

I'd aim to introduce the concepts, then illustrate with a practical example. This could perhaps be time-series based in order to emphasise the retrospective vs prospective aspects of intervention and counterfactuals.

Suggested categories:

• Level: Intermediate
• Diataxis type: Explanation or How-To

Related notebooks

This notebook would compliment the range of other causal focussed notebooks that I've put together:

It could be a good sequel to the Interventional distributions and graph mutation with the do-operator example, but it wouldn't really overlap with it.

References

• Pearl