Causal Inference in Python: Theory to Practice

A talk by Dr Dimitra Liotsiou from dunhumby.
Most data scientists know that ‘association does not imply causation’. However, traditional data science and machine learning methods are about association, not causation. At the same time, causal questions are central to many data science problems across sectors, e.g. questions about measuring effects, drivers, incrementally, or about why a change in a certain KPI took place. In this session, we will show how the recently developed mathematical apparatus for causal inference (graphical causal models and do-calculus) enables data scientists to move from association to causation, and we’ll demonstrate the application of the causal data science pipeline on a retail sector problem using the DoWhy library in Python.
You can find all relevant resources as referred to in this talk on our website:
datasciencefestival.com/sessi...

[@sroy2138]

This is a highly informative and useful presentation. It is clear, concise, and to the point.

[@DataScienceFestival]

Glad to hear it! 🎉

[@NugrohoBudianggoro]

bookmarking 23:08

[@anveshikakamble3717]

Without the data, I am unable to see any estimands. For all the 3 estimands it shows no such variables found. How can I know what variables to adjust ?

[@user-kr1no4eb7z]

Good challenge - you can try to create synthetic data (column names provided) based on your assumptions for distributions/rules and see what will happen ;)