Switchback Experiments under Geometric Mixing
Switchback Experiments under Geometric Mixing
复制标题
几何混合下之字形实验
DOI：
10.2139/ssrn.2606408
复制
发表时间：
2022
期刊：
影响因子：
--
通讯作者：
Stefan Wager
中科院分区：
文献类型：
--
作者：
Yuchen Hu;Stefan Wager
回切是一种实验设计，通过重复打开和关闭整个系统的干预来衡量治疗效果。回转实验是克服跨单元溢出效应的一种稳健方法；然而，它们容易受到时间结转的偏差的影响。在这篇文章中，我们考虑了以几何速率混合的马尔可夫系统的折返实验的性质。我们发现，在这种情况下，标准的回转设计严重受到结转偏差的影响：它们的估计误差在实验视界$T$中衰减为$T^{-1/3}$，而在没有结转的情况下，可能会有更快的速率$T^{-1/2}$。然而，我们还表明，明智地使用老化周期可以显著改善这种情况，并使错误衰减为$\log(T)^{1/2}T^{-1/2}$。我们的正式结果反映在经验评估中。
The switchback is an experimental design that measures treatment effects by repeatedly turning an intervention on and off for a whole system. Switchback experiments are a robust way to overcome cross-unit spillover effects; however, they are vulnerable to bias from temporal carryovers. In this paper, we consider properties of switchback experiments in Markovian systems that mix at a geometric rate. We find that, in this setting, standard switchback designs suffer considerably from carryover bias: Their estimation error decays as $T^{-1/3}$ in terms of the experiment horizon $T$, whereas in the absence of carryovers a faster rate of $T^{-1/2}$ would have been possible. We also show, however, that judicious use of burn-in periods can considerably improve the situation, and enables errors that decay as $\log(T)^{1/2}T^{-1/2}$. Our formal results are mirrored in an empirical evaluation.
DOI：
10.1287/opre.2021.2249
发表时间：
2019-09
期刊：
Oper. Res.
影响因子：
--
作者：
Nathan Kallus;Masatoshi Uehara
通讯作者：
Nathan Kallus;Masatoshi Uehara
具有时间干扰的自适应实验设计：最大似然法
DOI：
--
发表时间：
2020
期刊：
Advances in neural information processing systems
影响因子：
--
作者：
Glynn, Peter W;Johari, Ramesh;Rasouli, Mohammad
通讯作者：
Rasouli, Mohammad