scikit-learn · jeremiedbb · Apr 7, 2022 · Aug 31, 2020 · Aug 31, 2020 · Aug 31, 2020
diff --git a/doc/modules/partial_dependence.rst b/doc/modules/partial_dependence.rst
@@ -11,10 +11,10 @@ Partial dependence plots (PDP) and individual conditional expectation (ICE)
 plots can be used to visualize and analyze interaction between the target
 response [1]_ and a set of input features of interest.
 
-Both PDPs and ICEs assume that the input features of interest are independent
-from the complement features, and this assumption is often violated in practice.
-Thus, in the case of correlated features, we will create absurd data points to
-compute the PDP/ICE.
+Both PDPs [H2009]_ and ICEs [G2015]_ assume that the input features of interest
+are independent from the complement features, and this assumption is often
+violated in practice. Thus, in the case of correlated features, we will
+create absurd data points to compute the PDP/ICE [M2019]_.
 
 Partial dependence plots
 ========================
@@ -164,6 +164,18 @@ PDPs. They can be plotted together with
     ...     kind='both')
     <...>
 
+If there are too many lines in an ICE plot, it can be difficult to see
+differences between individual samples and interpret the model. Centering the
+ICE at the first value on the x-axis, produces centered Individual Conditional
+Expectation (cICE) plots [G2015]_. This puts emphasis on the divergence of
+individual conditional expectations from the mean line, thus making it easier
+to explore heterogeneous relationships. cICE plots can be plotted by setting
+`centered=True`:
+
+    >>> PartialDependenceDisplay.from_estimator(clf, X, features,
+    ...     kind='both', centered=True)
+    <...>
+
 Mathematical Definition
 =======================
 
@@ -255,15 +267,19 @@ estimators that support it, and 'brute' is used for the rest.
 
 .. topic:: References
 
-    T. Hastie, R. Tibshirani and J. Friedman, `The Elements of
-    Statistical Learning <https://web.stanford.edu/~hastie/ElemStatLearn//>`_,
-    Second Edition, Section 10.13.2, Springer, 2009.
-
-    C. Molnar, `Interpretable Machine Learning
-    <https://christophm.github.io/interpretable-ml-book/>`_, Section 5.1, 2019.
-
-    A. Goldstein, A. Kapelner, J. Bleich, and E. Pitkin, :arxiv:`Peeking Inside the
-    Black Box: Visualizing Statistical Learning With Plots of Individual
-    Conditional Expectation <1309.6392>`,
-    Journal of Computational and Graphical Statistics, 24(1): 44-65, Springer,
-    2015.
+    .. [H2009] T. Hastie, R. Tibshirani and J. Friedman,
+               `The Elements of Statistical Learning
+               <https://web.stanford.edu/~hastie/ElemStatLearn//>`_,
+               Second Edition, Section 10.13.2, Springer, 2009.
+
+    .. [M2019] C. Molnar,
+               `Interpretable Machine Learning
+               <https://christophm.github.io/interpretable-ml-book/>`_,
+               Section 5.1, 2019.
+
+    .. [G2015] :arxiv:`A. Goldstein, A. Kapelner, J. Bleich, and E. Pitkin,
+               "Peeking Inside the Black Box: Visualizing Statistical
+               Learning With Plots of Individual Conditional Expectation"
+               Journal of Computational and Graphical Statistics,
+               24(1): 44-65, Springer, 2015.
+               <1309.6392>`
diff --git a/doc/whats_new/v1.1.rst b/doc/whats_new/v1.1.rst
@@ -572,12 +572,19 @@ Changelog
   :pr:`16061` by `Thomas Fan`_.
 
 - |Enhancement| In
-  :meth:`~sklearn.inspection.PartialDependenceDisplay.from_estimator` and
-  :meth:`~sklearn.inspection.PartialDependenceDisplay.from_predictions`, allow
+  :meth:`inspection.PartialDependenceDisplay.from_estimator`, allow
   `kind` to accept a list of strings to specify  which type of
   plot to draw for each feature interaction.
   :pr:`19438` by :user:`Guillaume Lemaitre <glemaitre>`.
 
+- |Enhancement| :meth:`inspection.PartialDependenceDisplay.from_estimator`,
+  :meth:`inspection.PartialDependenceDisplay.plot`, and
+  :func:`inspection.plot_partial_dependence` now support plotting centered
+  Individual Conditional Expectation (cICE) and centered PDP curves controlled
+  by setting the parameter `centered`.
+  :pr:`18310` by :user:`Johannes Elfner <JoElfner>` and
+  :user:`Guillaume Lemaitre <glemaitre>`.
+
 :mod:`sklearn.isotonic`
 .......................
 

diff --git a/examples/inspection/plot_partial_dependence.py b/examples/inspection/plot_partial_dependence.py
@@ -113,7 +113,13 @@
 
 from sklearn.inspection import PartialDependenceDisplay
 
-common_params = {"subsample": 50, "n_jobs": 2, "grid_resolution": 20, "random_state": 0}
+common_params = {
+    "subsample": 50,
+    "n_jobs": 2,
+    "grid_resolution": 20,
+    "centered": True,
+    "random_state": 0,
+}
 
 print("Computing partial dependence plots...")
 tic = time()
@@ -188,10 +194,12 @@
 # rooms per household.
 #
 # The ICE curves (light blue lines) complement the analysis: we can see that
-# there are some exceptions, where the house price remain constant with median
-# income and average occupants. On the other hand, while the house age (top
-# right) does not have a strong influence on the median house price on average,
-# there seems to be a number of exceptions where the house price increase when
+# there are some exceptions (which are better highlighted with the option
+# `centered=True`), where the house price remains constant with respect to
+# median income and average occupants variations.
+# On the other hand, while the house age (top right) does not have a strong
+# influence on the median house price on average, there seems to be a number
+# of exceptions where the house price increases when
 # between the ages 15-25. Similar exceptions can be observed for the average
 # number of rooms (bottom left). Therefore, ICE plots show some individual
 # effect which are attenuated by taking the averages.