A new research paper introduces a framework for selecting online experiment designs when the mechanism of interference is unknown. The proposed method, called robust design selection, evaluates six different designs based on worst-case planning risk, considering factors like bias, variance, cost, and estimand mismatch. The paper provides theoretical guarantees and demonstrates its application on public datasets, recommending specific designs for Criteo ads, Open Bandit, and KuaiRand based on their respective risks. AI
RANK_REASON The cluster contains a research paper published on arXiv detailing a new framework for experiment design.
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
