pymc-labs Example Datasets
CausalPy ships with built-in datasets that can be loaded with cp.load_data(...).
Usage
import causalpy as cp
df = cp.load_data("did")
Available Datasets
| Key | Typical use | Description |
|---|---|---|
"did" |
Difference-in-differences | Synthetic DiD example data |
"banks" |
Difference-in-differences | Historic banking closures data |
"its" |
Interrupted time series | Seasonal synthetic ITS data |
"its simple" |
Interrupted time series | Simplified synthetic ITS data |
"covid" |
Interrupted time series | Deaths and temperature data for England and Wales |
"sc" |
Synthetic control | Synthetic control example data |
"brexit" |
Synthetic control | UK GDP data for Brexit causal impact |
"california_prop99" |
Synthetic control | California Proposition 99 cigarette sales panel |
"rd" |
Regression discontinuity | Synthetic RD example data |
"drinking" |
Regression discontinuity | Minimum legal drinking age data |
"geolift1" |
Geo experiments | Single-treatment geo-lift data |
"geolift_multi_cell" |
Geo experiments | Multi-cell geo-lift data |
"anova1" |
PrePostNEGD | Pre/post nonequivalent groups example |
"risk" |
Instrumental variables | Acemoglu, Johnson, and Robinson institutions data |
"schoolReturns" |
Instrumental variables | Schooling returns data |
"nhefs" |
Inverse propensity weighting | National Health and Nutrition Examination Survey data |
"lalonde" |
Inverse propensity weighting | LaLonde propensity-score data |
"nets" |
Inverse propensity weighting | National Supported Work Demonstration data |
"pisa18" |
General examples | PISA 2018 sample data |
"nevo" |
General examples | Berry, Levinsohn, and Pakes cereal data |
"zipcodes" |
Geo experiments | Zipcode-level geo-experiment data |
Guidance
- Prefer these bundled datasets for examples and docs instead of fetching data at runtime.
- For method selection, use
choosing-causalpy-methodsafter identifying the data shape. - For fitting and plotting, use
running-causalpy-experiments.