Reproduce “Task-specific experimental design for treatment effect estimation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Bethany Connolly, Kim Moore, Tobias Schwedes, Alexander Adam, Gary Willis, Ilya Feige, Christopher Frye
The authors develop a sampling method for targeted-marketing experiments that selects customer groups and treatment assignments according to how observations are expected to improve a later decision. A VAE latent grid, Bayesian outcome models, and task-specific scores guide sequential or batched sampling. On three marketing datasets, it generally matches RCT targeting performance with about one-tenth the sampled outcomes, while uniform latent-space sampling performs similarly.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.