Home > Events > 2017 Seminars & Colloquia > Lan Wang, University of Minnesota

Lan Wang, University of Minnesota

Main Content

Quantile-guided Optimal Policy Learning
When
05 April 2018 from 4:00 PM to 5:00 PM
Where
104 Thomas Building
Contact Name
Add event to calendar
vCal
iCal

Finding the optimal treatment assignment policies (sometimes called treatment regimes) or a series of sequential policies based on individual characteristics has important applications in areas such as government policies and active labor market interventions and precision medicine. In the current literature, the optimal treatment assignment policies is usually defined as the one that maximizes the average benefit in the potential population. We propose a general framework for estimating the quantile-optimal treatment assignment policies, which is of importance in many real-world applications. Given a collection of treatment assignment policies, we consider robust estimation of the quantile-optimal treatment assignment policies, which does not require the analyst to specify an outcome regression model. We propose an alternative formulation of the estimator as a solution of an optimization problem with an estimated nuisance parameter. This novel representation allows us to derive theory involving a nonstandard convergence rate and a non-normal limiting distribution. The same nonstandard convergence rate would also occur if the mean optimality criterion is applied, but this has not been studied. We will investigates both static and dynamic treatment assignment policies. In addition, doubly robust estimation and survival outcomes will be considered. (Joint work with Yu Zhou, Rui Song and Ben Sherwood)

Navigation for this Section

Events