Must-read papers and resources related to causal inference and machine (deep) learning
-
Updated
Nov 23, 2022
Must-read papers and resources related to causal inference and machine (deep) learning
train models in pytorch, Learn to Rank, Collaborative Filter, Heterogeneous Treatment Effect, Uplift Modeling, etc
Automated Double Machine Learning (DoubleML + AutoML) pipeline for causal inference — estimate incremental impact, CATE & ROI of promotions, pricing and media.
Interpretable and model-robust causal inference for heterogeneous treatment effects using generalized linear working models with targeted machine-learning
Code for TEDVAE, a VAE-based treatment effect estimation algorithm.
A Python Framework for Automatically Evaluating various Uplift Modeling Algorithms to Estimate Individual Treatment Effects
Implementation of Conformal Convolution T-learner (CCT) and Conformal Monte Carlo (CMC) learner
Package for heterogeneous treatment and spillover effects under network interference
Code for causal isotonic calibration for heterogeneous treatment effects (appeared in ICML, 2023)
Debiased Front-Door Learners for Heterogeneous Effects
ultimate matching toolbox & heterogenous treatment effects in panel data
Subgroup analyses: exploratory and subgroup identification
Code supplement for "Neuroevolutionary representations for learning heterogeneous treatment effects"
A Scala 3 ML Library for estimating causal effects
End-to-end, pre-registered uplift study on the Hillstrom e-mail RCT: CATE learners, budget-constrained allocation (integer programming + Lagrangian), and doubly robust policy evaluation. Includes a positive control, a heterogeneity * sample-size sweep, profit objectives, policy trees, chance constraints and a budgeted bandit. Fully reproducible.
Stata package for doubly robust uniform confidence bands for group-time conditional average treatment effects (CATT) in staggered difference-in-differences. Implements Imai, Qin, and Yanagi (2025, JBES).(Public Preview, testing...)(Disclaimer: CURRENTLY WIP)
4-arm randomized controlled A/B/n experiment evaluating curriculum modalities with Bonferroni-Holm adjustments and device HTE.
Personalized Treatment Evaluator
A modular Python benchmark for uplift modeling on the Criteo dataset, comparing S-Learner, T-Learner, X-Learner, DR-Learner, Causal Forest, and response-model targeting policies.
Double Descent and HTE
To associate your repository with the heterogeneous-treatment-effects topic, visit your repo's landing page and select "manage topics."