Finite Population Regression Adjustment and Non-asymptotic Guarantees for Treatment Effect Estimation

Authors: Mehrdad Ghadiri, David Arbour, Tung Mai, Cameron Musco, Anup B. Rao

NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We compliment the above theoretical bounds with experiments using both synthetic and real data, focusing on the full observation ATE and ITE estimation problems.
Researcher Affiliation Collaboration Mehrdad Ghadiri MIT mehrdadg@mit.edu David Arbour Adobe Research arbour@adobe.com Tung Mai Adobe Research tumai@adobe.com Cameron Musco UMass Amherst cmusco@cs.umass.edu Anup B. Rao Adobe Research anuprao@adobe.com
Pseudocode Yes Algorithm 1: ATE estimation with leverage score sampling and cross adjustment, Algorithm 2: (Ridge) leverage score sampling based regression adjustment, Algorithm 3: ITE estimation with random vector regression and leverage score sampling, Algorithm 4: Regression Adjusted Horvitz-Thompson Estimation, Algorithm 5: Non-Uniform Gram-Schmidt Walk
Open Source Code No The paper does not provide explicit statements about the release of its source code or links to a code repository for its methodology.
Open Datasets Yes 1. Boston Dataset [17]: This is a dataset of housing prices in the Boston area, consisting of 506 samples and 13 features. 2. IHDP Dataset [16, 11]: Derived from the characteristics of children and their mothers, this dataset comprises 747 samples and 25 features. 3. Twins Dataset [3]: This dataset is constructed based on the characteristics and mortality rates of twin births in the US.
Dataset Splits No The paper discusses the datasets used but does not provide specific details on training, validation, and test splits for reproducibility.
Hardware Specification No The paper does not provide specific details about the hardware used to run the experiments.
Software Dependencies No The paper does not provide specific software dependencies with version numbers.
Experiment Setup No The paper describes the generation of synthetic datasets and the number of trials for estimation, but it does not specify concrete hyperparameter values or detailed system-level training settings.