AExGym
收藏资源简介:
AExGym是由哥伦比亚大学开发的一个开源软件包,旨在为实验者提供一个易于评估和开发算法的平台。该数据集包含240个设置,涉及多个实际挑战,如批量反馈、非平稳性、多目标和结果等。数据集的创建过程结合了现实世界的数据,如ASOS.com数据集和多个现场实验数据。AExGym主要应用于改进静态实验设计,通过引入几轮适应性来提高实验效率,特别是在A/B测试领域。
AExGym is an open-source software package developed at Columbia University, designed to provide experimenters with a user-friendly platform for algorithm evaluation and development. This dataset includes 240 settings covering a range of real-world challenges such as batch feedback, non-stationarity, multi-objective scenarios, and outcome-related issues. The dataset is constructed by incorporating real-world data sources including the ASOS.com dataset and multiple field experimental datasets. AExGym is primarily applied to improve static experimental design, enhancing experimental efficiency by introducing several rounds of adaptivity, particularly in the domain of A/B testing.




