A Unified Workflow for Causal Inference.

Causal inference across binary, categorical, and continuous treatments — one consistent API. scikit-learn and skpro compatible.

For observational ML workflows

skcausal estimates causal effects from observational data. One fit(X, t, y) contract covers binary, categorical, and continuous treatments.

pip install skcausal
minimal continuous-treatment example 
from sklearn.linear_model import LinearRegression
from skcausal.causal_estimators import DirectRegressor

estimator = DirectRegressor(
    outcome_regressor=LinearRegression()
)
estimator.fit(X, t, y)
curve = estimator.predict(t)

Choose your path

Continuous treatments

Treatment is a dose, exposure, or intensity and the target is a dose-response curve. Includes density and GPS tooling.

Open continuous guide

Categorical treatments

Treatment has more than two named levels such as control, placebo, and treated.

Open categorical guide

Multi-dimensional treatments

Treatment has multiple dimensions (e.g. dose plus regime).

Open multi-dimensional guide