mcf – Modified Causal Forest in Python
mcf is a Python package implementing the Modified Causal Forest (MCF) methodology introduced by Lechner (2018) for estimating heterogeneous causal effects.
It provides a flexible framework for causal machine learning with support for binary and multiple treatments in both experimental and observational data settings.
In addition to treatment effect estimation, mcf enables data-driven policy learning through optimal treatment allocation rules based on estimated potential outcomes.
Documentation and Maintenance
- Documentation and website: https://mcfpy.github.io/mcf/#/
- Bug tracker: https://github.com/mcfpy/mcf/issues
Main Features
The mcf package provides two core components for causal machine learning:
Modified Causal Forest
The ModifiedCausalForest class implements a flexible tree-based framework for estimating heterogeneous treatment effects. It supports:
- Estimation of and inference for individualized treatment effects (IATEs) and aggregates as the (group) average treatment effects (GATEs, ATEs)
- Binary and multiple treatment settings
- Experimental and observational data
- Flexible covariate specification (ordered and unordered variables)
The object-oriented design provides a unified workflow for model training, prediction, and extraction of causal estimates.
Optimal Policy Learning
The OptimalPolicy class enables data-driven treatment assignment by learning decision rules that maximize a reward function. It provides:
- Learning of optimal treatment allocation rules
- Policy evaluation on training and test data
- Support for multiple treatment alternatives and policy constraints
This allows translation of policy scores into actionable decision rules for optimal policy design.
Citation
The implementation builds on the following literature:
- Lechner (2019): Modified Causal Forests for Estimating Heterogeneous Causal Effects. arXiv:1812.09487.
@misc{lechner2019modifiedcausalforest,
title = {Modified Causal Forests for Estimating Heterogeneous Causal Effects},
author = {Michael Lechner},
year = {2019},
eprint = {1812.09487},
archivePrefix = {arXiv},
primaryClass = {econ.EM},
url = {https://arxiv.org/abs/1812.09487}
}
License
This project is distributed under the MIT License.
4 maintainers
Metadata
4 maintainers
Release files for mcf 0.11.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | |
|---|---|---|---|---|
| mcf-0.11.0.tar.gz | 553.9 kB | Without provenance | Details |
Release files / mcf-0.11.0.tar.gz
| Download URL | mcf-0.11.0.tar.gz |
|---|---|
| Size | 553.9 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
1d9f423eeba7a013353c30878d901268c7c3e80100ec38043ecc1d1e4eef0f73
|
|
BLAKE2b-256 checksum How to use checksums |
e63987ce7375fac62721eb35a8568837cea784b38ebf523b363a26da1310ff0f
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.12.3
|
Security for mcf 0.11.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.
