Frengression

This repository contains code for our article, Frugal, Flexible, Faithful: Causal Data Simulation via Frengression.

Abstract

Machine learning has revitalized causal inference by combining flexible models and principled estimators, yet robust benchmarking and evaluation remain challenging with real-world data. In this work, we introduce frengression, a deep generative realization of the frugal parameterization that models the joint distribution of covariates, treatments and outcomes around the causal margin of interest. Frengression provides accurate estimation and flexible, faithful simulation of multivariate, time-varying data; it also enables direct sampling from user-specified interventional distributions. Model consistency and extrapolation guarantees are established, with validation on real-world clinical trial data demonstrating frengression’s practical utility. We envision this framework sparking new research into generative approaches for causal margin modelling.

Structure

Frengression model is stored at frengression.py. To replicate the experiments in the paper, please check the paper_exp folder for the jupyter notebooks.

Requirements

The R packages, causl and survivl need to be installed to run synthetic experiments. You can install the package via:

1. Install devtools

   install.packages("devtools")
   library(devtools)
   

2. Install causl and survivl

   install_github("rje42/causl")
   
   install_github("rje42/survivl")
   

Details of the packages can be found in https://github.com/rje42/causl and https://github.com/rje42/survivl.

The python package engression should be installed. Details can be found in https://github.com/xwshen51/engression.