What-if- Causal reasoning meets Bayesian Inference
summary
A Python package called CausalPy package implements quasi‑experimental designs and can fit either scikit‑learn estimators or PyMC Bayesian models; the demo builds a synthetic control for UK GDP and reports a cumulative Brexit impact with roughly 94% posterior mass around −3 to −5 billion USD. The code produces pre‑post fits (R² ≈ 0.96), plus interrupted time series, regression discontinuity and diff‑in‑diff utilities, and visualization of counterfactuals and credible intervals.
AI-written summary · about
chapters (31)
- 0:00Introduction to talk
- 0:31Package announcement
- 1:13Causal inference is trending
- 1:51Is it convincing enough?
- 2:41Bayesian model on the trends using PyMC
- 3:45Hype Cycle for emerging tech by Gartner
- 4:10Difference between Statistical relationships and Causal relationships
- 6:55Observational study on causal relationship between Tea and Death
- 9:00Confounding variables in our study
- 9:33Randomized control trial (RCT)
- 10:25Can you model confounding variables and not randomize?
- 12:17Randomization is very effective
- 12:35Randomized control trials can be problematic
- 14:56Quasi-Experimentation by Charles S. Reichardt
- 16:08CausalPy package
- 16:44What does CausalPy do?
- 19:00Normalized GDP
- 19:55What do we not have on this graph?
- 21:18Fitting the model
- 22:32Synthetic control method in CausalPy
- 23:28Visualizing the output
- 26:03Other features of CausalPy
- 26:52Regression discontinuity
- 27:49Difference in differences
- 28:04Did my advertising budget cause more sales?
- 29:41Summary
- 30:54Q/A Any suggested resource to properly design RCT?
- 31:56Q/A Why didn't you use a diff-diff model?
- 32:51Q/A Training a ML model to predict pre-treatment GDP of UK
- 34:07Q/A How is CausalPy related to CausalImpact
- 35:07Q/A Interrupted time series and regression discontinuity
related talks
24mAn introduction to Causal Inference with Python – making accurate estimates of cause and effect from
43mA Practical Guide to Causality in Python (For The Perplexed)
26mCausal inference with Causallib
31mIntroduction to Causal Inference using pgmpy
32mCausal inference and scenario generation within Just Eat Takeaway.com
30mClimbing the causal ladder for fun and profit
21mHazards on the Causal Path
59mA Welcome to Causal Probabilistic Programming