This repository contains a lightweight and bias-aware implementation of doubly robust Average Treatment Effect (ATE) wiht python code, including two most important doubly robust methods: Augmented Inverse Probability Weighting (AIPW) and Targeted Maximum Likelihood Estimation (TMLE).
Designed for clarity and reproducibility, the implementation only uses core scientific Python packages: pandas
, numpy
, and statsmodels
. It is especially suitable for researchers and practitioners who want interpretable and customizable code without relying on black-box machine learning libraries.
- ✅ Implements AIPW and TMLE with a focus on bias correction
- ✅ Supports both continuous and probability outcomes
- ✅ Minimal dependencies (
pandas
,numpy
,statsmodels
) - ✅ Clear and transparent logic for each step of the ATE pipeline
- Causal inference in observational studies
- Policy evaluation and impact analysis
- Education, public health, and social sciences research
Python_TMLE_AIPW.py
: Main script that prepares data, estimates nuisance functions, and computes ATE using both AIPW and TMLE.
- Python 3.8+
pandas
numpy
statsmodels