pyananke 0.6.1

Ananke, named for the Greek primordial goddess of necessity and causality, is a python package for causal inference using the language of graphical models.

pyananke

A modernized fork of ananke-causal with Python 3.10–3.12 compatibility fixes and z-identifiability support.

Installation

pip install pyananke

For graph visualization support:

pip install pyananke[viz]

What's new in 0.6.0

• Python 3.10–3.12 compatibility (numpy 2.x, pandas 2.x)
• New ananke.identification.idz module: z-identifiability decision procedure (Bareinboim & Pearl, 2012)
• New ananke.identification.ZID class: higher-level API for z-ID

Original package

This package is a modernized fork of ananke-causal by Rohit Bhattacharya et al. All original contributors are credited in AUTHORS.

DoWhy integration

This package is the backend for z-identifiability support in DoWhy:

pip install dowhy[zid]

Ananke

Visit the website to find out more.

Ananke, named for the Greek primordial goddess of necessity and causality, is a python package for causal inference using the language of graphical models

Contributors

• Rohit Bhattacharya
• Jaron Lee
• Razieh Nabi
• Preethi Prakash
• Ranjani Srinivasan

Interested contributors should check out the CONTRIBUTING.md for further details.

Installation

If graph visualization is not required then install via pip:

pip install ananke-causal

Alternatively, the package may be installed from gitlab by cloning and cd into the directory. Then, poetry (see https://python-poetry.org) can be used to install:

poetry install

Install with graph visualization

If graphing support is required, it is necessary to install graphviz.

Non M1 Mac instructions

Ubuntu:

sudo apt install graphviz libgraphviz-dev pkg-config

Mac (Homebrew):

brew install graphviz

Fedora:

sudo yum install graphviz

Once graphviz has been installed, then:

pip install ananke-causal[viz] # if pip is preferred

poetry install --extras viz # if poetry is preferred

M1 Mac specific instructions

If on M1 see this issue. The fix is to run the following before installing:

brew install graphviz
python -m pip install \
    --global-option=build_ext \
    --global-option="-I$(brew --prefix graphviz)/include/" \
    --global-option="-L$(brew --prefix graphviz)/lib/" \
    pygraphviz