Causal Machine Learning | Uplift Modeling | Heterogeneous Treatment Effects

Introduction

A comprehensive collection of approaches to estimate individualized treatment effects with a focus on machine learning, known as causal machine learning or uplift modeling.

Benchmark studies

• Devriendt, F., Moldovan, D., & Verbeke, W. (2018). A literature survey and experimental evaluation of the state-of-the-art in uplift modeling: a stepping stone toward the development of prescriptive analytics. Big Data, 6(1), 13–41. https://doi.org/10.1089/big.2017.0104
• Gubela, R. M., Bequé, A., Gebert, F., & Lessmann, S. (2019). Conversion uplift in e-commerce: A systematic benchmark of modeling strategies. International Journal of Information Technology & Decision Making, 18(3), 747–791.
• Künzel, S. R., Sekhon, J. S., Bickel, P. J., & Yu, B. (2019). Metalearners for estimating heterogeneous treatment effects using machine learning. Proceedings of the National Academy of Sciences, 116(10), 4156–4165.
• Powers, S., Qian, J., Jung, K., Schuler, A., Shah, N. H., Hastie, T., & Tibshirani, R. (2017). Some methods for heterogeneous treatment effect estimation in high-dimensions. CoRR, arXiv:1707.00102v1.
• Wendling, T., Jung, K., Callahan, A., Schuler, A., Shah, N. H., & Gallego, B. (2018). Comparing methods for estimation of heterogeneous treatment effects using observational data from health care databases. Statistics in Medicine, 37, 3309–3324. https://doi.org/10.1002/sim.7820