azcausal: Causal Inference in Python
====================================================================
Causal inference is an important component of the experiment evaluation. We highly recommend to have a look at the open-source
book: `Causal Inference for The Brave and True <https://matheusfacure.github.io/python-causality-handbook/landing-page.html>`_
Currently, azcausal provides two well-known and widely used causal inference methods: Difference-in-Difference (DID) and
Synthetic Difference-in-Difference (SDID). Moreover, error estimates via Placebo, Boostrap, or JackKnife are available.
.. _Installation:
Installation
********************************************************************************
To install the current release, please execute:
.. code:: bash
pip install git+https://github.com/amazon-science/azcausal.git
.. _Usage:
Usage
********************************************************************************
.. code:: python
from azcausal.core.error import JackKnife
from azcausal.core.panel import CausalPanel
from azcausal.data import CaliforniaProp99
from azcausal.estimators.panel.sdid import SDID
from azcausal.util import to_panels
# load an example data set with the columns Year, State, PacksPerCapita, treated.
df = CaliforniaProp99().df()
# create the panel data from the frame and define the causal types
data = to_panels(df, 'Year', 'State', ['PacksPerCapita', 'treated'])
ctypes = dict(outcome='PacksPerCapita', time='Year', unit='State', intervention='treated')
# initialize the panel
panel = CausalPanel(data).setup(**ctypes)
# initialize an estimator object, here synthetic difference in difference (sdid)
estimator = SDID()
# run the estimator
result = estimator.fit(panel)
# run the error validation method
estimator.error(result, JackKnife())
# plot the results
estimator.plot(result)
# print out information about the estimate
print(result.summary(title="CaliforniaProp99"))
.. code:: bash
╭──────────────────────────────────────────────────────────────────────────────╮
| CaliforniaProp99 |
├──────────────────────────────────────────────────────────────────────────────┤
| Panel |
| Time Periods: 31 (19/12) total (pre/post) |
| Units: 39 (38/1) total (contr/treat) |
├──────────────────────────────────────────────────────────────────────────────┤
| ATT |
| Effect (±SE): -15.60 (±2.9161) |
| Confidence Interval (95%): [-21.32 , -9.8884] (-) |
| Observed: 60.35 |
| Counter Factual: 75.95 |
├──────────────────────────────────────────────────────────────────────────────┤
| Percentage |
| Effect (±SE): -20.54 (±3.8393) |
| Confidence Interval (95%): [-28.07 , -13.02] (-) |
| Observed: 79.46 |
| Counter Factual: 100.00 |
├──────────────────────────────────────────────────────────────────────────────┤
| Cumulative |
| Effect (±SE): -187.25 (±34.99) |
| Confidence Interval (95%): [-255.83 , -118.66] (-) |
| Observed: 724.20 |
| Counter Factual: 911.45 |
╰──────────────────────────────────────────────────────────────────────────────╯
.. image:: docs/source/images/sdid.png
.. _Estimators:
Estimators
********************************************************************************
- **Difference-in-Difference (DID):** Simple implementation of the well-known Difference-in-Difference estimator.
- **Synthetic Difference-in-Difference (SDID):** Arkhangelsky, Dmitry Athey, Susan Hirshberg, David A. Imbens, Guido W. Wager, Stefan Synthetic Difference-in-Differences American Economic Review 111 12 4088-4118 2021 10.1257/aer.20190159 https://www.aeaweb.org/articles?id=10.1257/aer.20190159. Implementation based on https://synth-inference.github.io/synthdid/
.. _Contact:
Contact
********************************************************************************
Feel free to contact me if you have any questions:
| `Julian Blank <http://julianblank.com>`_ (blankjul [at] amazon.com)
| Amazon.com
| Applied Scientist, Amazon
| 410 Terry Ave N, Seattle 98109, WA.
Raw data
{
"_id": null,
"home_page": "https://github.com/amazon-science/azcausal",
"name": "azcausal",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.7",
"maintainer_email": null,
"keywords": "causality, inference",
"author": "Julian Blank",
"author_email": "blankjul@amazon.com",
"download_url": "https://files.pythonhosted.org/packages/35/d2/8141977f6a022d4619b4cc1bb9788e3436bcaff3f96ae573b0141531233b/azcausal-0.2.2.tar.gz",
"platform": "any",
"description": "azcausal: Causal Inference in Python\n====================================================================\n\nCausal inference is an important component of the experiment evaluation. We highly recommend to have a look at the open-source\nbook: `Causal Inference for The Brave and True <https://matheusfacure.github.io/python-causality-handbook/landing-page.html>`_\n\n\nCurrently, azcausal provides two well-known and widely used causal inference methods: Difference-in-Difference (DID) and\nSynthetic Difference-in-Difference (SDID). Moreover, error estimates via Placebo, Boostrap, or JackKnife are available.\n\n\n.. _Installation:\n\nInstallation\n********************************************************************************\n\n\nTo install the current release, please execute:\n\n.. code:: bash\n\n pip install git+https://github.com/amazon-science/azcausal.git\n\n\n.. _Usage:\n\nUsage\n********************************************************************************\n\n\n.. code:: python\n\n from azcausal.core.error import JackKnife\n from azcausal.core.panel import CausalPanel\n from azcausal.data import CaliforniaProp99\n from azcausal.estimators.panel.sdid import SDID\n from azcausal.util import to_panels\n\n\n # load an example data set with the columns Year, State, PacksPerCapita, treated.\n df = CaliforniaProp99().df()\n\n # create the panel data from the frame and define the causal types\n data = to_panels(df, 'Year', 'State', ['PacksPerCapita', 'treated'])\n ctypes = dict(outcome='PacksPerCapita', time='Year', unit='State', intervention='treated')\n\n # initialize the panel\n panel = CausalPanel(data).setup(**ctypes)\n\n # initialize an estimator object, here synthetic difference in difference (sdid)\n estimator = SDID()\n\n # run the estimator\n result = estimator.fit(panel)\n\n # run the error validation method\n estimator.error(result, JackKnife())\n\n # plot the results\n estimator.plot(result)\n\n # print out information about the estimate\n print(result.summary(title=\"CaliforniaProp99\"))\n\n\n.. code:: bash\n\n \u256d\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u256e\n | CaliforniaProp99 |\n \u251c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2524\n | Panel |\n | Time Periods: 31 (19/12) total (pre/post) |\n | Units: 39 (38/1) total (contr/treat) |\n \u251c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2524\n | ATT |\n | Effect (\u00b1SE): -15.60 (\u00b12.9161) |\n | Confidence Interval (95%): [-21.32 , -9.8884] (-) |\n | Observed: 60.35 |\n | Counter Factual: 75.95 |\n \u251c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2524\n | Percentage |\n | Effect (\u00b1SE): -20.54 (\u00b13.8393) |\n | Confidence Interval (95%): [-28.07 , -13.02] (-) |\n | Observed: 79.46 |\n | Counter Factual: 100.00 |\n \u251c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2524\n | Cumulative |\n | Effect (\u00b1SE): -187.25 (\u00b134.99) |\n | Confidence Interval (95%): [-255.83 , -118.66] (-) |\n | Observed: 724.20 |\n | Counter Factual: 911.45 |\n \u2570\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u256f\n\n.. image:: docs/source/images/sdid.png\n\n.. _Estimators:\n\nEstimators\n********************************************************************************\n\n\n- **Difference-in-Difference (DID):** Simple implementation of the well-known Difference-in-Difference estimator.\n- **Synthetic Difference-in-Difference (SDID):** Arkhangelsky, Dmitry Athey, Susan Hirshberg, David A. Imbens, Guido W. Wager, Stefan Synthetic Difference-in-Differences American Economic Review 111 12 4088-4118 2021 10.1257/aer.20190159 https://www.aeaweb.org/articles?id=10.1257/aer.20190159. Implementation based on https://synth-inference.github.io/synthdid/\n\n.. _Contact:\n\nContact\n********************************************************************************\n\nFeel free to contact me if you have any questions:\n\n| `Julian Blank <http://julianblank.com>`_ (blankjul [at] amazon.com)\n| Amazon.com\n| Applied Scientist, Amazon\n| 410 Terry Ave N, Seattle 98109, WA.\n\n\n",
"bugtrack_url": null,
"license": "Apache License 2.0",
"summary": "Casual Inference",
"version": "0.2.2",
"project_urls": {
"Homepage": "https://github.com/amazon-science/azcausal"
},
"split_keywords": [
"causality",
" inference"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "35d28141977f6a022d4619b4cc1bb9788e3436bcaff3f96ae573b0141531233b",
"md5": "c8594cadd5c6a6f0f52636d1d755cb3d",
"sha256": "0c0c72a88edd67a1341aa092ee8d817359d7500e72f999805904358c4f596e8b"
},
"downloads": -1,
"filename": "azcausal-0.2.2.tar.gz",
"has_sig": false,
"md5_digest": "c8594cadd5c6a6f0f52636d1d755cb3d",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.7",
"size": 247613,
"upload_time": "2024-03-28T23:07:08",
"upload_time_iso_8601": "2024-03-28T23:07:08.431155Z",
"url": "https://files.pythonhosted.org/packages/35/d2/8141977f6a022d4619b4cc1bb9788e3436bcaff3f96ae573b0141531233b/azcausal-0.2.2.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-03-28 23:07:08",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "amazon-science",
"github_project": "azcausal",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"lcname": "azcausal"
}
JSON
