CI - Libs for Python

The leading Python repositories for causal inference include:

Patrick Blöbaum: Performing Root Cause Analysis with DoWhy, a Causal Machine-Learning Library

• DoWhy: Effect Estimation, DoWhy provides a unified interface for various causal inference methods and is based on a language that combines graphical models and potential outcomes frameworks. It focuses on explicitly modeling and testing causal assumptions.
  • GH
  • Docs

Figure 1

Figure 2

Figure 3

• EconML: Also a Microsoft Research project, EconML (Automated Learning and Intelligence for Causation and Economics) is a library designed for estimating heterogeneous treatment effects using machine learning techniques. It is particularly useful in economics and other data-driven fields.
• CausalNex: Developed by McKinsey, CausalNex is a toolkit for causal reasoning with Bayesian networks. It simplifies the process of learning causal structures, incorporating domain expertise, and estimating intervention effects.
• causal-learn: This library, part of the PyWhy ecosystem along with DoWhy, focuses on causal discovery, offering various algorithms for learning causality from data.
• CImpact: A modular library for causal impact analysis, CImpact is designed for time-series data and supports multiple time series models, including TensorFlow and Prophet, to estimate the effect of a specific intervention over time.
• bnlearn: This Python package is used for causal discovery by learning the structure of Bayesian networks from data, as well as for parameter estimation and inference.