EconML

Why we included this project

EconML answers a question most A/B testing tools dodge: how much does a treatment help different people, not just the average. It grew out of Microsoft Research's ALICE project, pairing econometrics with machine learning methods like double machine learning that run on observational data, so you do not need a clean experiment to get causal estimates. The API stays consistent while you swap in different ML models for the nuisance and effect components, and the results keep their causal interpretation, often with valid confidence intervals. That makes it a practical fit for teams doing data-driven personalization or program evaluation, and for researchers who want more than aggregate A/B numbers.