Dowhy

New features and example notebooks, several bug fixes, and runtime improvements. Now compatible with Python 3.12.

* Python 3.12 support
* A new distribution change method that's more robust and converges faster (Multiply-robust causal change attribution, [Quintas-Martinez et al. (2024)](https://arxiv.org/abs/2404.08839))
* Support for effect estimation over time-series data ([Notebook](https://github.com/py-why/dowhy/blob/main/docs/source/example_notebooks/timeseries/effect_inference_timeseries_data.ipynb))
* New rank-based anomaly scorer
* New example notebook on sale attribution ([Notebook](https://github.com/py-why/dowhy/blob/main/docs/source/example_notebooks/sales_attribution_intervention.ipynb))
* New example notebook applying DoWhy for counterfactual fairness ([Notebook](https://github.com/py-why/dowhy/blob/main/docs/source/example_notebooks/counterfactual_fairness_dowhy.ipynb))
* Misc. updates to improve efficiency
* Ask queries about DoWhy using [Gurubase.io](https://gurubase.io/g/dowhy)

Contributors: bloebp, amit-sharma, kursataktas, vivianqin214, kapkic, GregVS, kmhj13, yangliu-SY, nparent1, rahulbshrestha, srivhash, darthtrevino, yogabonito, jonlives, krz, victor5as, sinhaharsh, Zethson, dw-610, diligejy

* New feature allowing users to write equations for the DGP of each node and obtain a causal model back with the mechanisms assigned (1106 )
* Convenience function to access fitted estimator instances from CausalModel (1113 )
* Bug fixes in Kernel-based independence test and networkx plot function
* Bug fixes for confidence intervals and regressionestimator
* Some improvements to CI/CD (auto-check readme on each PR, updated package publishing process, fix for timeout error)

Contributors: bhatt-priyadutt, drawlinson, bloebp, amit-sharma

* New functional API is ready for use. Try out the [notebook](https://www.pywhy.org/dowhy/v0.11/example_notebooks/dowhy_functional_api.html)
* A [notebook](https://www.pywhy.org/dowhy/v0.11/example_notebooks/dowhy_causal_discovery_example.html) showing how to use causal-learn graph discovery with DoWhy
* New [notebook](https://www.pywhy.org/dowhy/v0.11/example_notebooks/gcm_icc.html) demonstrating use of the intrinsic causal influence feature
* Enhanced compatibility between GCM and CausalModel api
* Frontdoor identification now supports multiple variables
* New [module](https://www.pywhy.org/dowhy/v0.11/user_guide/modeling_gcm/model_evaluation.html) for evaluating performance and falsifying assumptions of GCM models
* GCM auto assignment now returns a summary
* Extended documentation, revised and simpler README
* Bug fixes and improvements

A big thank you to all the contributors: amit-sharma, bloebp, kunwuz

This is a patch release.
* Added support for exposing interventional outcomes (drawlinson)
* Fixed bugs for pandas 2.0 support (bloebp) and confidence value for statistical test (amit-sharma)
* Additions to invariant nodes in GCM (bhatt-priyadutt)
* Fixing release pipeline (kbattocchi)

Thanks to everyone for contributing issues and fixes for this patch.

* Introducing an updated **[user guide](https://www.pywhy.org/dowhy/main/user_guide/intro.html)** for navigating the world of causality. The user guide is a great resource to learn about the different causal tasks, which ones may be relevant for you, and how to implement them using DoWhy.
* **Causal prediction** is the latest task supported by DoWhy! Try out the [prediction notebook](https://github.com/py-why/dowhy/blob/main/docs/source/example_notebooks/prediction/dowhy_causal_prediction_demo.ipynb) by jivatneet
* A new technique for **validating causal graphs**. Check out the [notebook](https://github.com/py-why/dowhy/blob/main/docs/source/example_notebooks/gcm_falsify_dag.ipynb) by eeulig
* **New refutation**: Overrule for learning boolean rules to describe support of the data/overlap between treatment and control groups in the data. Check out the [notebook](https://github.com/py-why/dowhy/blob/main/docs/source/example_notebooks/dowhy_refuter_assess_overlap.ipynb) by moberst
* Added a new method to estimate intrinsic causal influences for a single sample.
* Refactor of estimator API that allows separate fit and estimate methods
* Several optimizations and speed-ups of GCM methods
* Python 3.11 support and a simpler dependency list

A big thanks to all the contributors. AlxndrMlk amit-sharma andresmor-ms bloebp darthtrevino eeulig eltociear emrekiciman jivatneet kbattocchi Klesel MFreidank MichaelMarien moberst Padarn petergtz RoseDeSicilia26 sgrimbly vspinu yoshiakifukushima Zethson

Minor update to v0.9.

* Python 3.10 support
* Streamlined dependency structure for the dowhy package (fewer required dependencies)
* Color option for plots (eeulig)

Thanks darthtrevino, petergtz, andresmor-ms for driving this release!