Implementation:Scikit learn Scikit learn PartialDependenceDisplay

Knowledge Sources	Scikit_learn Scikit-learn Docs
Domains	Machine Learning, Model Inspection
Last Updated	2026-02-08 15:00 GMT

Overview

Concrete tool for visualizing Partial Dependence Plots (PDP) and Individual Conditional Expectations (ICE) provided by scikit-learn.

Description

The PartialDependenceDisplay class provides visualization of partial dependence and individual conditional expectation plots. It displays how one or two features marginally affect model predictions, supporting 1D line plots for single features (with optional ICE curves per sample) and 2D contour plots for feature interactions. The class is typically created via the from_estimator classmethod, which computes partial dependence internally using the partial_dependence function and arranges multiple subplots in a configurable grid layout.

Usage

Use this display class when interpreting black-box model predictions, understanding the marginal effect of features on model output, comparing the effect of different features, or visualizing individual conditional expectations alongside average partial dependence.

Code Reference

Source Location

Repository: scikit-learn
File: sklearn/inspection/_plot/partial_dependence.py

Signature

class PartialDependenceDisplay:
    def __init__(
        self,
        pd_results,
        *,
        features,
        feature_names,
        target_idx,
        deciles,
        kind="average",
        subsample=1000,
        random_state=None,
        is_categorical=None,
    )
    @classmethod
    def from_estimator(
        cls,
        estimator,
        X,
        features,
        *,
        sample_weight=None,
        categorical_features=None,
        feature_names=None,
        target=None,
        response_method="auto",
        n_cols=3,
        grid_resolution=100,
        percentiles=(0.05, 0.95),
        method="auto",
        n_jobs=None,
        verbose=0,
        line_kw=None,
        ice_lines_kw=None,
        pd_line_kw=None,
        contour_kw=None,
        ax=None,
        kind="average",
        subsample=1000,
        random_state=None,
        centered=False,
        custom_values=None,
    )
    def plot(self, *, ax=None, n_cols=3, line_kw=None, ice_lines_kw=None,
             pd_line_kw=None, contour_kw=None, bar_kw=None, heatmap_kw=None,
             pdp_lim=None, centered=False)

Import

from sklearn.inspection import PartialDependenceDisplay

I/O Contract

Inputs

Name	Type	Required	Description
pd_results	list of Bunch	Yes	Results of partial_dependence for each feature set
features	list of (int,) or list of (int, int)	Yes	Feature indices for each subplot
feature_names	list of str	Yes	Feature names corresponding to indices
target_idx	int	Yes	Target class index for multiclass or multioutput
deciles	dict	Yes	Decile values for each feature index
kind	str or list of str	No	Plot type: average (PDP), individual (ICE), or both (default average)
subsample	int	No	Number of ICE lines to display per plot (default 1000)
estimator	estimator instance	Yes (from_estimator)	Fitted estimator
X	array-like	Yes (from_estimator)	Input data for computing partial dependence
n_cols	int	No	Maximum number of columns in the subplot grid (default 3)
grid_resolution	int	No	Number of grid points per feature (default 100)
method	str	No	Computation method: auto, recursion, brute
centered	bool	No	Whether to center ICE/PD curves at first grid point (default False)

Outputs

Name	Type	Description
display	PartialDependenceDisplay	Display object with axes_, lines_, deciles_vlines_, figure_, and bounding_ax_ attributes

Usage Examples

Basic Usage

import matplotlib.pyplot as plt
from sklearn.datasets import make_friedman1
from sklearn.ensemble import GradientBoostingRegressor
from sklearn.inspection import PartialDependenceDisplay

X, y = make_friedman1(random_state=0)
est = GradientBoostingRegressor(n_estimators=100, random_state=0).fit(X, y)

# Plot partial dependence for features 0, 1, and interaction (0, 1)
features = [0, 1, (0, 1)]
PartialDependenceDisplay.from_estimator(
    est, X, features, kind="both", n_cols=3
)
plt.suptitle("Partial Dependence Plots")
plt.tight_layout()
plt.show()

# Plot ICE curves centered at the first grid point
PartialDependenceDisplay.from_estimator(
    est, X, features=[0, 1], kind="individual", centered=True
)
plt.show()

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment