causl 0.1.0

Unified CATE estimation: metalearners, neural nets, and boosted trees

causl

Unified CATE estimation in Python.

causl provides a clean, sklearn-compatible interface for heterogeneous treatment effect estimation, including metalearners, neural networks, and boosted tree methods.

Installation

pip install causl

Quickstart

from causl import SLearner, TLearner, DragonNet, NEDNet, CausalXGBoost
from sklearn.ensemble import GradientBoostingRegressor

# Metalearner
model = SLearner(base_learner=GradientBoostingRegressor())
model.fit(X, T, Y)
ite = model.predict_ite(X)
ate = model.predict_ate(X)

# Neural
model = DragonNet(input_dim=X.shape[1])
model.fit(X, T, Y)

Models

Model	Type	Class
S-Learner	Metalearner	SLearner
T-Learner	Metalearner	TLearner
DragonNet	Neural	DragonNet
NEDNet	Neural	NEDNet
CXGBoost	Tree-based	CausalXGBoost

License

MIT