The MCF package provides a powerful Python program that allows you to deploy the Modified Causal Forest for your application. The scope of the MCF is to estimate heterogeneous treatment effects for multiple treatment models in a selection-on-observables setting.
This website documents the package in detail, explaining how the programme is structured, which parameters can be specified and how this impacts estimation and running times.
To get started, check out the How to install guide. To perform your first estimation, start here. Further, we provide example scripts at different levels of complexity in the directory examples. The data is provided here. For a thorough discussion read through the MCF Core Walkthrough. Here, every programme step is briefly outlined giving you a deeper understanding of the implementation. The Python API provides a comprehensive list of the functional inputs. For the succinct version consult the Short Python API. For technical details consult Lechner (2019). Bodory, Busshoff and Lechner (2022) showcase the package replicating studies in the fields of epidemiology, medicine, and labor economics.
The release notes can be found here.
The MCF package has been made with ♥ in Sankt Gallen, Switzerland! Support us by getting involved as hinted here.