UpSet Plot

This example demonstrates the UpSet plot for visualizing feature interactions. The UpSet plot shows the most important interactions as vertical bars with the interacting features in a matrix below.

from __future__ import annotations

from sklearn.model_selection import train_test_split
from xgboost import XGBRegressor

import shapiq

Train a Model and Compute Explanations

x_data, y_data = shapiq.datasets.load_california_housing(to_numpy=False)
feature_names = list(x_data.columns)
x_data, y_data = x_data.values, y_data.values
x_train, x_test, y_train, y_test = train_test_split(
    x_data,
    y_data,
    test_size=0.2,
    random_state=42,
)
model = XGBRegressor(random_state=42, max_depth=4, n_estimators=50)
model.fit(x_train, y_train)

x_explain = x_test[2]
explainer = shapiq.TabularExplainer(
    model,
    data=x_test,
    index="FSII",
    max_order=2,
    random_state=42,
)
explanation = explainer.explain(x_explain, budget=200)
print(explanation)

InteractionValues(
    index=FSII, max_order=2, min_order=0, estimated=True, estimation_budget=200,
    n_players=8, baseline_value=2.058321475982666,
    Top 10 interactions:
        (): 2.058321475982666
        (7,): 1.3230916479551968
        (5,): 0.41396584141388837
        (6, 7): 0.31170752719138173
        (1, 7): 0.267773292536492
        (2, 5): 0.21734101617495935
        (1,): 0.14959330449384228
        (1, 5): 0.1408693332978724
        (0,): -0.2003426692827929
        (6,): -0.5329538181989749
)

Basic UpSet Plot

explanation.plot_upset()

Customization: Number of Interactions

explanation.plot_upset(n_interactions=10)

Color the Matrix

Color-code the matrix based on interaction sign (red = positive, blue = negative).

explanation.plot_upset(color_matrix=True)

Feature Names

explanation.plot_upset(feature_names=feature_names, n_interactions=15)

Show Only Relevant Features

Set all_features=False to show only features present in the top interactions.

explanation.plot_upset(all_features=False, n_interactions=7, feature_names=feature_names)

Adjust Figure Size

explanation.plot_upset(n_interactions=5, figsize=(5, None))