/causalGPNs

Causal inference from observational data with Gaussian process networks.

Primary LanguageR

This repository contains all the code to reproduce the results in Bayesian Causal Inference with Gaussian Process Networks.

  • /A_Thaliana contains the data, code and results for the application to the A. Thaliana dataset.
  • /Local_approx contains the implementation and simulation of the local approximation.
  • /Simulation_plots contains all generated plots from the simulation study.

The .Stan files contain code for the Gaussian (Gauss.stan) and additive GP (Add.stan) models.

  • BayesStanFns.R and GPscore.R contain functions to compute the score according to the GP and GP^2 models respectively.
  • Fourier_fns.R contains functions to generate non-linear data for the simulations.
  • Fxsampling_fns.R contains the functions to implement Bayesian causal inference with the GPN model.
  • LinGaussian_estimate.R generates the simulation results for the linear-Gaussian model for causal inference from observational data.
  • MC_estimate.R generates the simulation results for causal inference with the GPN model.

Reference

@misc{giudice2024bayesian,
      title={Bayesian Causal Inference with Gaussian Process Networks}, 
      author={Enrico Giudice and Jack Kuipers and Giusi Moffa},
      year={2024},
      eprint={2402.00623},
      archivePrefix={arXiv},
      primaryClass={stat.ML}
}