Skip to content

namkoong-lab/aexgym

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

AExGym: Benchmarks and Environments for Adaptive Experimentation

About

AExGym includes benchmark environments for adaptive experimentation based on real- world datasets, highlighting prominent practical challenges to operationalizing adaptivity: non-stationarity, batched/delayed feedback, multiple outcomes and objectives, and external validity. Our benchmark aims to spur methodological development that puts practical performance (e.g., robustness) as a central concern, rather than mathematical guarantees on contrived instances.AExGym is designed with modularity and extensibility in mind to allow experimentation practitioners to develop and benchmark custom environments and algorithms.

Installation

To install dependencies for this project with anaconda, use:

conda env create -f environment.yml
conda activate aexgym

Environments

We construct environments from the following datasets:

  • ASOS Digital Experiments Dataset link: A set of 78 online experiments with multiple metrics conducted by ASOS.com, an international fashion retailer.
  • Pennsylvania Reemployment Bonus Demonstration link: A large study of the effectiveness of different bonus amounts in accelerating reemployment and reducing reliance on unemployment insurance.
  • Meager Multi-Site Study of Microcredit Expansions link: A multi-site study of Randomized Control Trials that investigates the impacts of microcredit expansions across seven countries.
  • National Health Interview Survey link: A survey on a broad range of health topics are collected through personal household interviews.

Run Example Environments

The notebooks folder contains the code examples for the environments described in our paper. Each example contains instructions on how to download and process the datasets used to construct the environment. Datasets go in the data folder.

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published