Skip to content

Latest commit

 

History

History
28 lines (23 loc) · 2.25 KB

libraries.md

File metadata and controls

28 lines (23 loc) · 2.25 KB
layout title description
page
Python Libraries for CausalML
A collection of python libraries that are useful for causal machine learning.

A collection of python libraries that are useful for causal machine learning.

General

  • PyWhy. an organization that holds a multitude of causal machine learning packages in python
  • DoWhy. A package that offers a high level API for graph definition, identification, estimation, sensitivity analysis and refutation
  • PGMPY. A package for general Bayesian network definition and inference (not just causal inference)

Identification

  • Ananke. A package that implements advanced graph based identification algorithms and allows for semi-parametric estimation.
  • y0. Advanced graph based identification algorithms for general causal graphs (even allowing for un-directed edges)

Estimation

  • EconML. Many causal ML algorithms for estimation and confidence intervals for heterogeneous treatment effects (under conditional exogeneity or with instruments)
  • CausalML. Many causal ml algorithms for estimation of heterogeneous treatment effects
  • UpliftML. A smaller set of causal ML algorithms for heterogeneous effect estimation, but which scales on Spark, using PySpark
  • DoubleML. Implements the double ML estimation algorithm with inference, for average treatment effects under exogeneity or with instruments

Causal Graph Discovery

  • PyWhy-Graphs. A basic library for causal graph manipulation and basic graph algorithms (e.g. conditional independence)
  • DoDiscover. A package for causal discovery in python
  • Causica. A deep learning based causal discovery package
  • CausalLearn. Causal discovery package with many recent advanced algorithms from the research community