libraries.md

Causal Inference Libraries

The following is a list of causal inference libraries, ordered by language and popularity (stars).

• Python
• R
• Julia

Python

• Statsmodels

  • regression - Linear regression models
  • tsa - Time series analysis
  • duration - Survival and duration models
  • nonparametric - Nonparametric methods
  • gmm - Generalized method of moments
  • stats - Statistical tests and tools

• Linearmodels (extension of Statsmodels)

  • iv - Instrumental variables
  • panel - Panel data models
  • system - System regression models (e.g. seemingly unrelated regressions)

• DoWhy

  • gcm - Graphical causal model-based inference

• CausalML

  • inference.tree - Tree-based uplift models [Docs]
  • inference.meta - S-, T-, X-, R-, DR-, TMLE-learners [Docs]
  • inference.iv - Doubly-robust instrumental variables [Docs]
  • match - Matching
  • propensity - Propensity score estimation

• EconML

  • dml - Double machine learning
  • dr - Doubly robust learning
  • forest - Causal forests
  • metalearners - S-, T-, X-learners [Paper]
  • iv.dml - Double machine learning with instrumental variables
  • iv.nnet - Deep instrumental variables
  • dynamic_dml - Dynamic double machine learning

• CausalNex

  • structure - Directed acyclic graph structure learning
  • network - Bayesian network modeling
  • evaluation - Model evaluation
  • inference - Model inference

• causal-learn

  • search - Search methods for causal discovery

• CausalPy

  • InterruptedTimeSeries - Interrupted time series
  • SyntheticControl - Synthetic control methods
  • DifferenceInDifferences - Difference-in-differences
  • RegressionDiscontinuity - Regression discontinuity

• GLuM

  • Fast generalized linear models [Docs]

• causal-tune

  • AutoML for causal inference

• GeoLift

  • Uplift modeling for geographical experiments [Docs]

• pyhdfe

  • Fast high-dimensional fixed effect orthogonalization algorithm

• trimmed_match

  • Trimmed match estimator for aggregate geographical experiments [Paper]

• pyfixest

  • Fast high-dimensional fixed effect regression for ols, iv and poisson regression following fixest syntax
  • Difference-in-differences with staggered adaption: Gardner's 2-stage estimator ("did2s") and Dube et al's (2023) local projections approach (vignette link)

• DoubleML - Double/debiased machine learning [Paper] [Video]

• DeepIV - Deep learning for instrumental variables estimation [Paper]

• ananke - Causal inference with DAGs

• scikit-uplift - Basic meta-learner and uplift tools

R

• CausalImpact

  • Bayesian structural time-series models [Docs]

• Robyn

  • Marketing mixed models [Docs]

• grf

  • Generalized random forests [Docs]

• FixEst

  • Fast high-dimensional fixed effect regression [Docs]

• tidyhte

  • Tidy-style heterogeneous treatment effect estimation [Docs]

• alpaca - Generalized linear models with high-dimensional fixed effects

• DAGitty - [Video]

• synthdid - Synthetic difference in differences [Paper]

• ggdag - Visualizing and analyzing causal DAGs

• pcalg - Causal structure learning and causal inference using DAGs

Julia

• Tidier

  • Tidyverse port to Julia

• FixedEffectModels

  • Fast high-dimensional fixed effect regression

• Causalinference

  • Causal inference with directed acyclic graphs