arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2403.04039cs.LGstat.MEstat.ML
收藏

K臂随机实验下条件反事实均值估计的样本量规划

Sample size planning for conditional counterfactual mean estimation with a K-armed randomized experiment

Adobe Inc.(奥多比公司)

Gabriel Ruiz

AI总结:

针对K臂随机实验中数据驱动子群体的条件反事实均值估计问题,提出将样本量规划转化为同时推断问题的方法,可推导对应样本量或反向求解可行的处理臂数与划分复杂度,并在公开数据集上验证了性能保证。

摘要:

本文研究如何确定K臂随机实验的足够大样本量,以估计数据驱动子群体中的条件反事实期望。这些子群体可由任意特征空间划分算法输出,包括通过对具有相似预测分数的用户进行分箱所定义的子群体,或是由学习得到的策略树(policy tree)所定义的子群体。在明确设定推断目标、最低置信水平和最大误差边际后,核心思路是将原始目标转化为一个同时推断问题:为抵消估计误差上升的可能性,推荐的样本量与待执行的推断数量直接相关。在样本量预算固定的情况下,我们的结果可将问题反向转化为求解可行的处理臂数量或划分复杂度(例如决策树叶子节点数量)。我们利用策略树学习子群体,在一个大型公开随机实验测试数据集上评估了所提方法的名义保证性能。

英文摘要:

We cover how to determine a sufficiently large sample size for a $K$-armed randomized experiment in order to estimate conditional counterfactual expectations in data-driven subgroups. The sub-groups can be output by any feature space partitioning algorithm, including as defined by binning users having similar predictive scores or as defined by a learned policy tree. After carefully specifying the inference target, a minimum confidence level, and a maximum margin of error, the key is to turn the original goal into a simultaneous inference problem where the recommended sample size to offset an increased possibility of estimation error is directly related to the number of inferences to be conducted. Given a fixed sample size budget, our result allows us to invert the question to one about the feasible number of treatment arms or partition complexity (e.g. number of decision tree leaves). Using policy trees to learn sub-groups, we evaluate our nominal guarantees on a large publicly-available randomized experiment test data set.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑