Masahiro Kato's genriesz automates debiased ML for causal inference in Python
New Python package implements generalized Riesz regression to automatically estimate causal effects like ATE with confidence intervals.
Researcher Masahiro Kato released genriesz, an open-source Python package that automates debiased machine learning (DML) using generalized Riesz regression. It implements a unified framework for estimating Riesz representers via Bregman divergences, includes automatic regressor balancing (ARB), and provides modular interfaces for specifying target functionals and model classes. Users can estimate causal parameters like average treatment effects (ATE) and get regression adjustment, Riesz weighting, and TMLE-style estimators with cross-fitting and confidence intervals.
Why It Matters
Streamlines complex causal inference workflows for data scientists and econometricians, reducing manual implementation errors.