majiayu000 Running Placebo Analysis
Executes placebo-in-time sensitivity analysis to validate causal experiments.
Workflow
- Define Experiment Factory: Create a function that returns a fitted CausalPy experiment (e.g., ITS, DiD, SC) given a dataset and time boundaries.
- Configure Analysis: Initialize
PlaceboAnalysiswith the factory, dataset, intervention dates, and number of folds (cuts). - Run Analysis: Execute
.run()to fit models on pre-intervention data folds. - Evaluate Results: Compare placebo effects (which should be null) to the actual intervention effect. Use histograms and hierarchical models to quantify the "status quo" distribution.
Key Concepts
- Placebo-in-time: Simulating an intervention at a time when none occurred to check if the model falsely detects an effect.
- Fold: A slice of pre-intervention data used to test a placebo period.
- Factory Pattern: Decouples the placebo logic from the specific CausalPy experiment type.
References
- Placebo-in-time Implementation: Full code for the
PlaceboAnalysisclass, usage examples, and hierarchical status-quo modeling.