Causal Inference в ML: инструменты
Для тех, кто не читал статью на Хабре (3/3)

Tool Boxes для Python:

- Dowhy - Propensity-based Stratification, PSM, IPW, Regression
- Causal ML - Tree-based algorithms, X/T/X/R-learner
- CausalNex - Structural Causal Models based on Bayesian Networks
- EconML - Doubly Robust Learner, Orthogonal Random Forests, Meta-Learners, Deep Instrumental Variables
- causalImpact - Bayesian structural time-series model (сейчас активна реализация c бекендом на tensorflow-probability вместо pystan)

Tool Boxes для R:

- causalToolbox - BART, Causal Forest, T/X/S-learner with BART/RF as base learner
- causalImpact - Bayesian structural time-series model
- did - Classical Difference-in-Difference (group-time average treatment effects)
- synthdid - Synthetic difference in difference estimator (SDID) for the average treatment effect in panel data, Arkhangelsky et al (2019) – доклад на Causal Inference in ML Track 2020
- causalweight - Inverse probability weighting (IPW)

Если вы считаете, что стоит пополнить этот список - пишите в комментариях!

#tech #causal_inference