pyfixest 0.60.0

Fast high dimensional fixed effect estimation following syntax of the fixest R package.

PyFixest: Fast High-Dimensional Fixed Effects Regression in Python

Docs · Quickstart · Function & API Reference · DeepWiki · Benchmarks · Contributing · Changelog

PyFixest is a Python package for fast high-dimensional fixed effects regression.

The package aims to mimic the syntax and functionality of Laurent Bergé's formidable fixest package as closely as Python allows. If you know fixest well, the goal is that you won't have to read the docs to get started! In particular, this means that all of fixest's defaults are mirrored by PyFixest.

For questions on PyFixest, head over to our GitHub discussions, or join our Discord server.

Features

• Estimation
  • OLS, WLS, IV, and GLMs (Poisson, logit, probit, gaussian) with high-dimensional fixed effects
  • Different demeaning backends (MAP, within LSMR, torch LSMR) on CPU and GPU
  • Fast quantile regression via an interior-point solver
  • Difference-in-differences estimators, including TWFE, Did2s, local projections, and Sun-Abraham event studies
  • Regression decomposition following Gelbach (2016)
  • Multiple estimation syntax
• Inference
  • Several robust, cluster-robust, and HAC variance-covariance estimators
  • Wild cluster bootstrap inference via wildboottest
  • Multiple hypothesis corrections and simultaneous confidence intervals
  • Fast randomization inference
  • The causal cluster variance estimator (CCV)
• Post-Estimation & Reporting
  • Publication-ready tables with Great Tables or LaTeX booktabs via the maketables library

Installation

You can install the release version from PyPI by running

# inside an active virtual environment
python -m pip install pyfixest

or the development version from github by running

python -m pip install git+https://github.com/py-econometrics/pyfixest

Optional dependencies

For visualization features using the lets-plot backend, install:

python -m pip install pyfixest[plots]

matplotlib is included by default, so plotting works without this extra.

To run the LSMR demeaner via PyTorch (CPU and GPU), you need to install PyTorch, which you can do via

python -m pip install pyfixest[torch]

For GPU acceleration on CUDA, you additionally need to install a CUDA-enabled torch build. See the PyTorch installation guide for details.

Then use the typed demeaner API:

# CPU
pf.feols(
    "Y ~ X1 | f1 + f2",
    data=data,
    demeaner=pf.LsmrDemeaner(backend="torch", device="cpu"),
)

# CUDA GPU
pf.feols(
    "Y ~ X1 | f1 + f2",
    data=data,
    demeaner=pf.LsmrDemeaner(backend="torch", device="cuda"),
)

Note: The cupy / scipy LSMR backends and the jax MAP backend are deprecated and will be removed in a future release. Use pf.LsmrDemeaner() for the default Rust within LSMR backend on CPU, or the torch-based backend for GPU acceleration.

## Quickstart

import pyfixest as pf

data = pf.get_data()
pf.feols("Y ~ X1 | f1 + f2", data=data).summary()

###

Estimation:  OLS
Dep. var.: Y, Fixed effects: f1+f2
Inference:  CRV1
Observations:  997

| Coefficient   |   Estimate |   Std. Error |   t value |   Pr(>|t|) |   2.5% |   97.5% |
|:--------------|-----------:|-------------:|----------:|-----------:|-------:|--------:|
| X1            |     -0.919 |        0.065 |   -14.057 |      0.000 | -1.053 |  -0.786 |
---
RMSE: 1.441   R2: 0.609   R2 Within: 0.2

PyFixest also supports multiple estimation syntax:

fit = pf.feols("Y + Y2 ~ X1 | csw0(f1, f2)", data=data, vcov={"CRV1": "group_id"})
fit.etable()

For more examples, see the quickstart, the formula syntax tutorial, and the Poisson & GLMs tutorial.

Benchmarks

The DGPs follow the "simple" and "difficult" designs from the fixest benchmarks. The figure timings for regressions with k=10 covariates and plots the median runtime across three runs for PyFixest MAP, PyFixest within, PyFixest torch on CUDA GPU, fixest, and FixedEffectModels.jl.

Runtime comparisons are hardware-dependent, especially for GPU backends. For the full benchmark suite, see the benchmarks/ directory and the note on difficult fixed effects problems.

Learn More

• Quickstart
• Function & API Reference
• Difference-in-Differences
• Quantile Regression
• Changelog
• Contributing

Acknowledgements

First and foremost, we want to acknowledge Laurent Bergé's formidable fixest, which is so good we decided to stick to its API and conventions as closely as Python allows. Without fixest, PyFixest likely wouldn't exist - or at the very least, it would look very different.

For a full list of software packages and papers that have influenced PyFixest, please take a look at the Acknowledgements page.

We thank all institutions that have funded or supported work on PyFixest!

How to Cite

If you want to cite PyFixest, you can use the following BibTeX entry:

@software{pyfixest,
  author  = {{The PyFixest Authors}},
  title   = {{pyfixest: Fast high-dimensional fixed effect estimation in Python}},
  year    = {2025},
  url     = {https://github.com/py-econometrics/pyfixest}
}

Support PyFixest

If you enjoy using PyFixest, please consider donating to GiveDirectly and dedicating your donation to [e-mail]. You can also leave a message through the donation form; your support and encouragement mean a lot to the developers.

Call for Contributions

Thanks for showing interest in contributing to pyfixest! We appreciate all contributions and constructive feedback, whether that be reporting bugs, requesting new features, or suggesting improvements to documentation.

If you'd like to get involved, but are not yet sure how, please feel free to send us an email. Some familiarity with either Python or econometrics will help, but you really don't need to be a numpy core developer or have published in Econometrica =) We'd be more than happy to invest time to help you get started!

Contributors ✨

Thanks goes to these wonderful people:

styfenschaer 💻	Niall Keleher 🚇 💻	Wenzhi Ding 💻	Apoorva Lal 💻 🐛	Juan Orduz 🚇 💻	Alexander Fischer 💻 🚇	aeturrell ✅ 📖 📣	leostimpfle 💻 🐛
baggiponte 📖	Sanskriti 🚇	Jaehyung 💻	Alex 📖	Hayden Freedman 💻 📖	Aziz Mamatov 💻	rafimikail 💻	Benjamin Knight 💻
Dirk Sliwka 💻 📖	daltonm-bls 🐛	Marc-André 💻 🐛	Kyle F Butts 🔣	Marco Edward Gorelli 👀	Vincent Arel-Bundock 💻	IshwaraHegde97 💻	Tobias Schmidt 📖
escherpf 🐛 💻	Iván Higuera Mendieta 💻	Ádám Vig 💻	Szymon Sacher 💻	AronNemeth 💻	Dmitri Tchebotarev 💻	FuZhiyu 🐛 💻	Marcelo Ortiz M. 📖
Joseph Stover 📖	JaapCTJ 💻	Matt Shapiro 💻	Kristof Schröder 💻	Wiktor 💻	Daman Dhaliwal 💻	Jaakko Markkanen 🐛	Jonas Skjold Raaschou-Pedersen 💻 📖
Bobby Ho 📖	Erica Ryan 💻	Souhil Abdelmalek Louddad 📖	Demetri Pananos 💻 👀	Patrick Doupe 💻	Ariadna Albors Zumel 💻	Martien Lubberink 💻	jonpcohen 📖
rhstanton 📓 🐛	mdizadi 📖	Tim Mensinger 🚇	Nicholas Junge 💻 🚇	Abel Abate 💻	Jan Teusen (né Boelts) 💻	Marie Garkavenko 👀	Brad Hackinen 💻
Julian Budde 🐛	Felix Schmitz 💻

This project follows the all-contributors specification. Contributions of any kind welcome!