y0
(pronounced "why not?") is Python code for causal inference.
y0
has a fully featured internal domain specific language for representing
probability expressions:
from y0.dsl import P, A, B
# The probability of A given B
expr_1 = P(A | B)
# The probability of A given not B
expr_2 = P(A | ~B)
# The joint probability of A and B
expr_3 = P(A, B)
It can also be used to manipulate expressions:
from y0.dsl import P, A, B, Sum
P(A, B).marginalize(A) == Sum[A](P(A, B))
P(A, B).conditional(A) == P(A, B) / Sum[A](P(A, B))
DSL objects can be converted into strings with str()
and parsed back
using y0.parser.parse_y0()
.
A full demo of the DSL can be found in this Jupyter Notebook
y0
has a notion of acyclic directed mixed graphs built on top of
networkx
that can be used to model causality:
from y0.graph import NxMixedGraph
from y0.dsl import X, Y, Z1, Z2
# Example from:
# J. Pearl and D. Mackenzie (2018)
# The Book of Why: The New Science of Cause and Effect.
# Basic Books, p. 240.
napkin = NxMixedGraph.from_edges(
directed=[
(Z2, Z1),
(Z1, X),
(X, Y),
],
undirected=[
(Z2, X),
(Z2, Y),
],
)
y0
has many pre-written examples in y0.examples
from Pearl, Shpitser,
Bareinboim, and others.
y0
provides actual implementations of many algorithms that have remained
unimplemented for the last 15 years of publications including:
Algorithm | Reference |
---|---|
ID | Shpitser and Pearl, 2006 |
IDC | Shpitser and Pearl, 2008 |
ID* | Shpitser and Pearl, 2012 |
IDC* | Shpitser and Pearl, 2012 |
Surrogate Outcomes | Tikka and Karvanen, 2018 |
Apply an algorithm to an ADMG and a causal query to generate an estimand represented in the DSL like:
from y0.dsl import P, X, Y
from y0.examples import napkin
from y0.algorithm.identify import Identification, identify
# TODO after ID* and IDC* are done, we'll update this interface
query = Identification.from_expression(graph=napkin, query=P(Y @ X))
estimand = identify(query)
assert estimand == P(Y @ X)
The most recent release can be installed from PyPI with:
$ pip install y0
The most recent code and data can be installed directly from GitHub with:
$ pip install git+https://github.com/y0-causal-inference/y0.git
Contributions, whether filing an issue, making a pull request, or forking, are appreciated. See CONTRIBUTING.md for more information on getting involved.
The code in this package is licensed under the BSD-3-Clause license.
Before we publish an application note on y0
, you can cite this software
via our Zenodo record (also see the badge above):
@software{y0,
author = {Charles Tapley Hoyt and
Jeremy Zucker and
Marc-Antoine Parent},
title = {y0-causal-inference/y0},
month = jun,
year = 2021,
publisher = {Zenodo},
version = {v0.1.0},
doi = {10.5281/zenodo.4950768},
url = {https://doi.org/10.5281/zenodo.4950768}
}
This project has been supported by several organizations (in alphabetical order):
- Harvard Program in Therapeutic Science - Laboratory of Systems Pharmacology
- Pacific Northwest National Laboratory
The development of the Y0 Causal Inference Engine has been funded by the following grants:
Funding Body | Program | Grant |
---|---|---|
DARPA | Automating Scientific Knowledge Extraction (ASKE) | HR00111990009 |
PNNL Data Model Convergence Initiative | Causal Inference and Machine Learning Methods for Analysis of Security Constrained Unit Commitment (SCY0) | 90001 |
DARPA | Automating Scientific Knowledge Extraction and Modeling (ASKEM) | HR00112220036 |
This package was created with @audreyfeldroy's cookiecutter package using @cthoyt's cookiecutter-snekpack template.
See developer instructions
The final section of the README is for if you want to get involved by making a code contribution.
To install in development mode, use the following:
git clone git+https://github.com/y0-causal-inference/y0.git
cd y0
pip install -e .
This project uses cruft
to keep boilerplate (i.e., configuration, contribution guidelines, documentation
configuration)
up-to-date with the upstream cookiecutter package. Update with the following:
pip install cruft
cruft update
More info on Cruft's update command is available here.
After cloning the repository and installing tox
with pip install tox tox-uv
,
the unit tests in the tests/
folder can be run reproducibly with:
tox -e py
Additionally, these tests are automatically re-run with each commit in a GitHub Action.
The documentation can be built locally using the following:
git clone git+https://github.com/y0-causal-inference/y0.git
cd y0
tox -e docs
open docs/build/html/index.html
The documentation automatically installs the package as well as the docs
extra specified in the pyproject.toml
. sphinx
plugins
like texext
can be added there. Additionally, they need to be added to the
extensions
list in docs/source/conf.py
.
The documentation can be deployed to ReadTheDocs using
this guide.
The .readthedocs.yml
YAML file contains all the configuration you'll need.
You can also set up continuous integration on GitHub to check not only that
Sphinx can build the documentation in an isolated environment (i.e., with tox -e docs-test
)
but also that ReadTheDocs can build it too.
- Log in to ReadTheDocs with your GitHub account to install the integration at https://readthedocs.org/accounts/login/?next=/dashboard/
- Import your project by navigating to https://readthedocs.org/dashboard/import then clicking the plus icon next to your repository
- You can rename the repository on the next screen using a more stylized name (i.e., with spaces and capital letters)
- Click next, and you're good to go!
Zenodo is a long-term archival system that assigns a DOI to each release of your package.
- Log in to Zenodo via GitHub with this link: https://zenodo.org/oauth/login/github/?next=%2F. This brings you to a page that lists all of your organizations and asks you to approve installing the Zenodo app on GitHub. Click "grant" next to any organizations you want to enable the integration for, then click the big green "approve" button. This step only needs to be done once.
- Navigate to https://zenodo.org/account/settings/github/, which lists all of your GitHub repositories (both in your username and any organizations you enabled). Click the on/off toggle for any relevant repositories. When you make a new repository, you'll have to come back to this
After these steps, you're ready to go! After you make "release" on GitHub (steps for this are below), you can navigate to https://zenodo.org/account/settings/github/repository/y0-causal-inference/y0 to see the DOI for the release and link to the Zenodo record for it.
You only have to do the following steps once.
- Register for an account on the Python Package Index (PyPI)
- Navigate to https://pypi.org/manage/account and make sure you have verified your email address. A verification email might not have been sent by default, so you might have to click the "options" dropdown next to your address to get to the "re-send verification email" button
- 2-Factor authentication is required for PyPI since the end of 2023 (see this blog post from PyPI). This means you have to first issue account recovery codes, then set up 2-factor authentication
- Issue an API token from https://pypi.org/manage/account/token
You have to do the following steps once per machine. Create a file in your home directory called
.pypirc
and include the following:
[distutils]
index-servers =
pypi
testpypi
[pypi]
username = __token__
password = <the API token you just got>
# This block is optional in case you want to be able to make test releases to the Test PyPI server
[testpypi]
repository = https://test.pypi.org/legacy/
username = __token__
password = <an API token from test PyPI>
Note that since PyPI is requiring token-based authentication, we use __token__
as the user, verbatim.
If you already have a .pypirc
file with a [distutils]
section, just make sure that there is an index-servers
key and that pypi
is in its associated list. More information on configuring the .pypirc
file can
be found here.
After installing the package in development mode and installing
tox
with pip install tox tox-uv
,
run the following from the shell:
tox -e finish
This script does the following:
- Uses Bump2Version to switch the version number in
the
pyproject.toml
,CITATION.cff
,src/y0/version.py
, anddocs/source/conf.py
to not have the-dev
suffix - Packages the code in both a tar archive and a wheel using
build
- Uploads to PyPI using
twine
. - Push to GitHub. You'll need to make a release going with the commit where the version was bumped.
- Bump the version to the next patch. If you made big changes and want to bump the version by minor, you can
use
tox -e bumpversion -- minor
after.
- Navigate to https://github.com/y0-causal-inference/y0/releases/new to draft a new release
- Click the "Choose a Tag" dropdown and select the tag corresponding to the release you just made
- Click the "Generate Release Notes" button to get a quick outline of recent changes. Modify the title and description as you see fit
- Click the big green "Publish Release" button
This will trigger Zenodo to assign a DOI to your release as well.