Proximal Learning for Individualized Treatment Regimes Under Unmeasured Confounding
python >= 3.8
numpy >= 1.20
scipy >= 1.6.2
pandas >= 1.2.3
scikit-learn >= 0.24.1
pytorch >= 1.8.1
conda install pandas scikit-learn numpy scipy pytorch torchvision torchaudio cudatoolkit=11.1 -c pytorch -c conda-forge
- notebooks: Usage examples for all simulation settings in the paper, in each example:
samp_size
: sample size is 2000 (users can try other settings)qtl
: the quantile for dPESS selection, default is 0.4 (users can try other settings)
- data contains file to generate simulated data
- src source files:
- proxITR.py: main file of proximal ITR learning
- rkhs_scaler.py: estimators of ourcome bridge function h0 and treatment bridge function q0
- torchSVC.py: optimizer of weighted binary support vector classification
@article{qi2022proximal,
title={Proximal learning for individualized treatment regimes under unmeasured confounding},
author={Qi, Zhengling and Miao, Rui and Zhang, Xiaoke},
journal={Journal of the American Statistical Association},
number={just-accepted},
pages={1--33},
year={2022},
publisher={Taylor \& Francis}
}