genriesz: Generalized Riesz Regression

genriesz is a Python package for Automatic Debiased Machine Learning (ADML) with Generalized Riesz Regression (GRR) under Bregman divergences, with automatic regressor balancing.

The code is available on https://github.com/MasaKat0/genriesz.

The central user-facing API is:

• genriesz.grr_functional() for a generic linear functional,

• genriesz.grr_ate() for the ATE,

• genriesz.grr_att() for the ATT,

• genriesz.grr_did() for panel DID (implemented as ATT on ΔY),

• genriesz.grr_ame() for average marginal effects.

Contents

• Getting started
  • Installation
  • Notes
  • Quickstart: ATE
  • Next steps
• User guide
  • Conceptual procedure
  • Bases
  • Density ratio estimation
  • Generators and automatic links
  • Estimators, cross fitting, and outcome models
  • Regularization: \(\ell_p\)
  • Diagnostics and balance checks
• API reference
  • High-level estimation
  • Density ratio and covariate shift
  • Functionals
  • Low-level solvers
  • Bases
  • Generators
  • Matching helpers
  • Results
• Examples
  • Jupyter notebooks
  • Python scripts
• References
  • Primary references