Useful resources on causality
- 1 minRecognising true causal relationships is crucial in shaping effective interventions in medicine, policy-making, and business. Often randomised control trials aren’t possible, and we need to deduce causality just from observed data. In such cases, the widely known principle that correlation doesn’t imply causation comes to life.
Below is a collection of resources to learn more about causality.
Causality 101
If you have enough time on your hands, you can go directly to Judea Pearl’s seminal Causality: Models, Reasoning, and Inference. It’s quite heavy on theory but very comprehensive. For a more relaxed reading, you can check out The Book of Why by Judea Pearl and Dana Mackenzie.
A nice overview which gives a bird-eye view on causality in ML, and raises some hard open questions.
Bayesian Networks
DeepMind’s blog on Bayesian Networks
Software
- DoWhy: a Python library for Causal Inference by Microsoft
- Causal ML: a Python library for Uplift Modelling and Causal Inference by Uber
- CausalNex: a Python library that uses Bayesian Networks to combine machine learning and domain expertise for causal reasoning
Viktoriia Oliinyk
Building Responsible AI