software
Selected software and code accompanying my methodological work.
This page highlights selected research software accompanying my work in causal inference, semiparametric statistics, and calibrated machine learning. Each package is tied to a methodological paper and is intended to make the underlying ideas easier to inspect, reproduce, and use.
causalCalibration
causalCalibration accompanies Causal isotonic calibration for heterogeneous treatment effects. It provides post-hoc calibration and cross-calibration for black-box heterogeneous treatment effect estimators, improving reliability while preserving the original learning pipeline.
hte3
hte3 accompanies Combining T-learning and DR-learning: a framework for oracle-efficient estimation of causal contrasts. It implements efficient plug-in learning for heterogeneous causal contrasts, including the conditional average treatment effect and conditional relative risk.
ppi-aipw
ppi-aipw accompanies Prediction-Powered Inference via Calibration. It provides semisupervised mean inference with AIPW-style estimators and calibrated prediction scores for settings with limited labels and abundant unlabeled covariates.
calibratedDML
calibratedDML accompanies Doubly robust inference via calibration. It adds a calibration step to standard debiased machine learning pipelines for inference on linear functionals, with the goal of improving robustness and finite-sample performance.