Data

Tools

Indexes

Applications

Experiments

Open Source Science

https://github.com/aiml-k/activebayesiancausal

Code for ABC3: Active Bayesian Causal Inference with Cohn Criteria in Randomized Experiments (AAAI 2025)

https://github.com/aiml-k/activebayesiancausal

Science Score: 23.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ○
    CITATION.cff file
  • ✓
    codemeta.json file
    Found codemeta.json file
  • ○
    .zenodo.json file
  • ○
    DOI references
  • ✓
    Academic publication links
    Links to: arxiv.org
  • ○
    Academic email domains
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (8.4%) to scientific vocabulary
Last synced: 9 months ago · JSON representation

Repository

Code for ABC3: Active Bayesian Causal Inference with Cohn Criteria in Randomized Experiments (AAAI 2025)

Basic Info
  • Host: GitHub
  • Owner: AIML-K
  • License: mit
  • Language: Python
  • Default Branch: main
  • Size: 2.27 MB
Statistics
  • Stars: 2
  • Watchers: 1
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Created over 1 year ago · Last pushed over 1 year ago
Metadata Files
Readme License

README.md

ABC3: Active Bayesian Causal Inference with Cohn Criteria in Randomized Experiments (AAAI 2025)

This is a code for ABC3: Active Bayesian Causal Inference with Cohn Criteria in Randomized Experiments.

Paper

Abstract

In causal inference, a randomized experiment is a de facto method to overcome various theoretical issues in observational study. However, the experimental design requires expensive costs, so an efficient experimental design is necessary. We propose ABC3, a Bayesian active learning policy for causal inference. We show a policy minimizing an estimation error on conditional average treatment effect is equivalent to minimizing an integrated posterior variance, similar to Cohn criteria. We theoretically prove ABC3 also minimizes an imbalance between the treatment and control groups and the type 1 error probability. Imbalance-minimizing characteristic is especially notable as several works have emphasized the importance of achieving balance. Through extensive experiments on real-world data sets, ABC3 achieves the highest efficiency, while empirically showing the theoretical results hold.

Installation

  1. Install the requirements by pip install -r requirements.txt

  2. Install torch following the instruction

Main Experiments

  1. Run the run.sh by bash run.sh

  2. Check the resulting plots in plots/cate_error, plots/mmd and plots/type1

Other Experiments

  • If you are interested in reproducing the hyperparameter test (Fig. 4 and 5), run run/hyperparameter_kernel.py and hyperparameter_sigma.py with desired arguments, then run src/plot.py with arguments kernel or sigma.

  • If you want to check our assumption (Fig. 6), run run/assumption.py, then run src/plot.py wirh a assumption argument.

  • If you want to test our sampling-and-optimization-based algorithm (Appendix E), run run/sampling.py with proper arguments, then run src/plot.py with a sampling argument.

  • If you want to plug-in different regressors (Appendix F), run run/regressor.py with proper arguments, then run src/plot.py with a reg arguement.

Owner

  • Name: AIML-K
  • Login: AIML-K
  • Kind: organization

AI+Math Lab @ Korea Univ.

GitHub Events

Total
  • Watch event: 7
  • Member event: 2
  • Push event: 1
  • Create event: 4
Last Year
  • Watch event: 7
  • Member event: 2
  • Push event: 1
  • Create event: 4

Dependencies

requirements.txt pypi
  • matplotlib *
  • numpy *
  • pandas *
  • scikit-learn *
  • scipy *
  • tqdm *