Causal Methods implemented in Python.
Install via
pip install causalipy
It might make sense to add the py-arrow dependency (which is currently required for the example):
pip install pyarrow
To run a version of the multi-period difference-in-difference estimator as
proposed by Callaway and Sant’Anna (2020) (this requires additionally pyarrow - e.g. via
pip install pyarrow
- to be installed currently):
from causalipy.did.multi_periods import MultiPeriodDid
import pandas as pd
url = "https://github.com/mohelm/causalipy-datasets/raw/main/mpdta-sample.feather"
data = pd.read_feather(url)
mpd_minimum_wage = MultiPeriodDid(
data,
outcome="lemp",
treatment_indicator="treat",
time_period_indicator="year",
group_indiciator="first.treat",
formula="~ 1",
)
mpd_minimum_wage.plot_treatment_effects()
This will give:
This project is licensed under the terms of the MIT license.