[ENH] Add the classifier CI test (py-why#28)

* Adding classifier CI test (ccit) and unit test * Adds a simulation module for simulating non-linear additive noise models * Add flaky to the unit test suite CCIT * Address typing issues with regards to adding sklearn and pytorch NN modules as the "classifier" Signed-off-by: Adam Li <adam2392@gmail.com>
ebridge2 · Sep 22, 2022 · 0771188 · 0771188
1 parent 5b7465b
commit 0771188
Show file tree

Hide file tree

Showing 11 changed files with 715 additions and 84 deletions.
diff --git a/doc/api.rst b/doc/api.rst
@@ -29,6 +29,7 @@ learning.
    KernelCITest
    GSquareCITest
    FisherZCITest
+   ClassifierCITest
 
 Graph protocols
 ===============

diff --git a/doc/conf.py b/doc/conf.py
@@ -116,6 +116,9 @@
     "node",
     "nodes",
     "graph",
+    # numpy
+    "ScalarType",
+    "ArrayLike",
     # shapes
     "n_times",
     "obj",
@@ -128,6 +131,9 @@
     "n_variables",
     "n_classes",
     "NDArray",
+    # deep learning
+    "pytorch",
+    "model",
 }
 numpydoc_xref_aliases = {
     # Networkx
@@ -142,11 +148,14 @@
     "DAG": "dodiscover.DAG",
     "BaseConditionalIndependenceTest": "dodiscover.ci.BaseConditionalIndependenceTest",
     "Context": "dodiscover.Context",
+    "Column": "dodiscover.typing.Column",
     # joblib
     "joblib.Parallel": "joblib.Parallel",
+    # numpy
+    "NDArray": "numpy.ndarray",
+    "ArrayLike": ":term:`array_like`",
     # pandas
     "pd.DataFrame": "pandas.DataFrame",
-    "pandas.DataFrame": "pandas.DataFrame",
     "column": "pandas.DataFrame.columns",
 }
 
@@ -175,14 +184,15 @@
 
 intersphinx_mapping = {
     "python": ("https://docs.python.org/3", None),
-    "numpy": ("https://numpy.org/devdocs", None),
+    "numpy": ("https://numpy.org/doc/stable/", None),
     "scipy": ("https://scipy.github.io/devdocs", None),
     "pandas": ("https://pandas.pydata.org/pandas-docs/dev", None),
     "networkx": ("https://networkx.org/documentation/latest/", None),
     "pgmpy": ("https://pgmpy.org", None),
     "sklearn": ("https://scikit-learn.org/stable", None),
     "joblib": ("https://joblib.readthedocs.io/en/latest", None),
     "matplotlib": ("https://matplotlib.org/stable", None),
+    "torch": ("https://pytorch.org/docs/master/", None),
 }
 intersphinx_timeout = 5
 

diff --git a/doc/references.bib b/doc/references.bib
@@ -16,6 +16,13 @@ @article{Colombo2012
   url       = {https://doi.org/10.1214/11-AOS940}
 }
 
+@article{Lopez2016revisiting,
+  title   = {Revisiting classifier two-sample tests},
+  author  = {Lopez-Paz, David and Oquab, Maxime},
+  journal = {arXiv preprint arXiv:1610.06545},
+  year    = {2016}
+}
+
 @article{Meek1995,
   author  = {Meek, Christopher},
   year    = {2013},
@@ -37,6 +44,14 @@ @book{Neapolitan2003
   doi       = {10.1145/1327942.1327961}
 }
 
+@article{Sen2017model,
+  title   = {Model-powered conditional independence test},
+  author  = {Sen, Rajat and Suresh, Ananda Theertha and Shanmugam, Karthikeyan and Dimakis, Alexandros G and Shakkottai, Sanjay},
+  journal = {Advances in neural information processing systems},
+  volume  = {30},
+  year    = {2017}
+}
+
 @article{Zhang2008,
   title    = {On the completeness of orientation rules for causal discovery in the presence of latent confounders and selection bias},
   journal  = {Artificial Intelligence},

diff --git a/doc/whats_new/v0.1.rst b/doc/whats_new/v0.1.rst
@@ -30,6 +30,7 @@ Changelog
 - |Feature| Implement conditional independence tests under the ``dodiscover.ci`` submodule, by `Adam Li`_ (:pr:`16`)
 - |Feature| Implement skeleton learning method, :class:`dodiscover.constraint.LearnSkeleton` under the ``dodiscover.constraint`` submodule, by `Adam Li`_ (:pr:`20`)
 - |Feature| Implement confusion matrix method for comparing networkx-like graphs, :func:`dodiscover.metrics.confusion_matrix_networks`, by `Adam Li`_ (:pr:`48`)
+- |Feature| Implement classification-based CI test (CCIT), :class:`dodiscover.ci.ClassifierCITest` under the ``dodiscover.ci`` submodule, by `Adam Li`_ (:pr:`28`)
 
 
 Code and Documentation Contributors

diff --git a/dodiscover/ci/__init__.py b/dodiscover/ci/__init__.py
@@ -1,4 +1,5 @@
 from .base import BaseConditionalIndependenceTest
+from .clf_test import ClassifierCITest
 from .fisher_z_test import FisherZCITest
 from .g_test import GSquareCITest
 from .kernel_test import KernelCITest