Alternative Approach to Causal Libraries

Having recently done a review of libraries on causality, I have the impression that Microsoft’s option is the most viable among others, since it has technical support, promotion and community support. However, I am not mostly a big fan of this “specify all the parameters you can and step aside” approach.

So I was surprised to find at one of the conferences a plausible alternative approach, which the authors themselves call “Causal probabilistic programming without tears”. Unlike learners, this py-library allows you to customize a lot of things related to estimation and tests and control the process to a much greater extent, so for anyone with an econometric or statistical background I highly recommend it.

You can find details here if you want to get to know it better:

• Paper
• GitHub Repository
• Documentation