CausalML / DTR-Bandit

Replication code for "DTR Bandit: Learning to Make Response-Adaptive Decisions With Low Regret" by Hu and Kallus

• https://arxiv.org/abs/2005.02791

Synthetic experiments

• Run Simulation/simulation.py to get reg_dtr.npy, reg_eps.npy, reg_lsvi.npy

• Run the Simulation/plot_*.py to get plots

STAR*D

The data is proprietary and cannot be provided. The data may be requested here https://nda.nih.gov/edit_collection.html?id=2148 by following the instructions here https://nda.nih.gov/training/module?trainingModuleId=training.access&slideId=slide.creating.package