Add gcm recipe to README (#431)
This commit is contained in:
Родитель
6a23076578
Коммит
249d8ecaef
20
README.rst
20
README.rst
|
@ -204,7 +204,25 @@ GCM-based inference (experimental)
|
||||||
----------------------------------
|
----------------------------------
|
||||||
|
|
||||||
Graphical causal model-based inference, or GCM-based inference for short, is an experimental addition to DoWhy. For
|
Graphical causal model-based inference, or GCM-based inference for short, is an experimental addition to DoWhy. For
|
||||||
details, check out the `documentation for the gcm sub-package <https://py-why.github.io/dowhy/gcm>`_.
|
details, check out the `documentation for the gcm sub-package <https://py-why.github.io/dowhy/gcm>`_. The basic
|
||||||
|
recipe for this API works as follows:
|
||||||
|
|
||||||
|
.. code:: python
|
||||||
|
|
||||||
|
# 1. Modeling cause-effect relationships as a structural causal model
|
||||||
|
# (causal graph + functional causal models):
|
||||||
|
scm = gcm.StructuralCausalModel(nx.DiGraph([('X', 'Y'), ('Y', 'Z')])) # X -> Y -> Z
|
||||||
|
scm.set_causal_mechanism('X', gcm.EmpiricalDistribution())
|
||||||
|
scm.set_causal_mechanism('Y', gcm.AdditiveNoiseModel(gcm.ml.create_linear_regressor()))
|
||||||
|
scm.set_causal_mechanism('Z', gcm.AdditiveNoiseModel(gcm.ml.create_linear_regressor()))
|
||||||
|
|
||||||
|
# 2. Fitting the SCM to the data:
|
||||||
|
gcm.fit(scm, data)
|
||||||
|
|
||||||
|
# 3. Answering a causal query based on the SCM:
|
||||||
|
results = gcm.<causal_query>(scm, ...)
|
||||||
|
|
||||||
|
Where <causal_query> can be one of multiple functions explained in `Answering Causal Questions <https://py-why.github.io/dowhy/gcm/user_guide/answering_causal_questions/index.html>`_.
|
||||||
|
|
||||||
|
|
||||||
A high-level Pandas API
|
A high-level Pandas API
|
||||||
|
|
Загрузка…
Ссылка в новой задаче