Unified Covariate Adjustment for Causal Inference

Authors: Yonghan Jung, Jin Tian, Elias Bareinboim

NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments corroborate the scalability and robustness of the proposed framework. We demonstrate scalability and robustness to bias both theoretically and empirically through simulations.
Researcher Affiliation Academia Yonghan Jung1, Jin Tian2, and Elias Bareinboim3 1Purdue University 2Mohamed bin Zayed University of Artificial Intelligence 3Columbia University
Pseudocode Yes Algorithm 1: Tian-to-UCA(G, V := (V1, , VK, Y )) Algorithm 2: DML-UCA({Di}, L)
Open Source Code No The paper states in the NeurIPS checklist that 'Every detail in the simulations is sufficiently provided to reproduce the results' but does not provide a direct link to a code repository or explicitly state that the code is open-source.
Open Datasets No The paper states in Appendix F: 'We define the following structural causal models:... We include a segment of the code employed to generate the dataset.' This indicates synthetic data generation rather than the use of a publicly available dataset with concrete access information.
Dataset Splits Yes (Sample splitting) For each i [m + 1], randomly split Di iid P i into L-folds. Let Di ℓdenote the ℓ-th partition, and define Di ℓ:= Di \ Di ℓ.
Hardware Specification No Appendix F, in the XGBoost parameterization section, mentions 'n_jobs: 4 # Assuming you have 4 cores', but no specific hardware models (CPU, GPU, or cloud instances) are provided.
Software Dependencies No Appendix F states: 'As described in Sec. 4, we used the XGBoost (Chen and Guestrin, 2016) as a model for estimating nuisances. We implemented the model using Python.' No specific version numbers for Python or XGBoost are given.
Experiment Setup Yes Appendix F details parameterizations for XGBoost, including 'mu_params' and 'pi_params' dictionaries with specific values for 'booster', 'eta', 'gamma', 'max_depth', 'min_child_weight', 'subsample', 'colsample_bytree', 'lambda', 'alpha', 'objective', 'eval_metric', and 'n_jobs'.