eduzhai > Applied Sciences > Computer Science >

Double/Debiased Machine Learning for Dynamic Treatment Effects

  • KanKan
  • (0) Download
  • 20210424
  • Save

... pages left unread,continue reading

Document pages: 28 pages

Abstract: We consider the estimation of treatment effects in settings when multipletreatments are assigned over time and treatments can have a causal effect onfuture outcomes. We formulate the problem as a linear state space Markovprocess with a high dimensional state and propose an extension of thedouble debiased machine learning framework to estimate the dynamic effects oftreatments. Our method allows the use of arbitrary machine learning methods tocontrol for the high dimensional state, subject to a mean square errorguarantee, while still allowing parametric estimation and construction ofconfidence intervals for the dynamic treatment effect parameters of interest.Our method is based on a sequential regression peeling process, which we showcan be equivalently interpreted as a Neyman orthogonal moment estimator. Thisallows us to show root-n asymptotic normality of the estimated causal effects.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×