A introduction to causal inference using common tools from the python data stack
Thank you for signing up for the Causal Inference tutorial!
I am confident that by the end of this session you will:
- Understand the pitfalls of observational data analysis
- Know the various types of causal relationships to look out for
- Describe the hierarchy of statistical analyses, causal inference, and experiments
- Start conducting preliminary causal analyses on your own data
- Confidently explore the topic on your own (now that you have a solid foundational understanding of causal thinking)
If you have any suggestions or questions, I would love to hear them (even outside the SciPy 2023 conference). My email is roni.kobrosly@gmail.com.
The only technical requirements for this tutorial are having a local machine with an internet connection, a common web browser (e.g. Firefox, Safari, Chrome, etc.), and Google account (for accessing Colab and Google Slides).
All exercise notebooks (both "student" and "teacher" versions) are available on Google Colab. Notebooks have been thoroughly tested and can run end-to-end in the provided environments. See links below.
The notebooks uses datasets which are publicly available in this GitHub repo. Don't worry, though, the Colab notebooks directly link to them and you don't have to do anything beyond run the Colab notebook code as is. In case you want to play with the datasets locally (beyond the tutorial notebooks), they are conveniently included in this tutorial repo in the data folder.
Slides to go along with the tutorial session are available as a Google Doc Presentation.
- Notebook 1 - Causal Graphs:
- Notebook 2 - G-computation/S-learner:
- Notebook 3 - Causal Curves / Continuous Treatments: