Implementation:Interpretml Interpret Explain Local

Field	Value
Sources	InterpretML
Domains	Interpretability, Visualization
Last Updated	2026-02-07 12:00 GMT

Overview

Explain_Local is a concrete tool for generating per-sample local explanations provided by the InterpretML EBM classes.

Description

The explain_local method decomposes predictions for specific samples into per-term contributions. For each sample, it evaluates each term's score tensor at the sample's bin indices, producing a horizontal bar chart showing how each feature contributed to the prediction. It also includes the intercept, predicted value, and optional actual value.

Usage

Call this method on a fitted EBM with specific samples to explain individual predictions.

Code Reference

Source Location

Repository: interpretml/interpret
File: python/interpret-core/interpret/glassbox/_ebm/_ebm.py
Lines: 2326--2466

Signature

def explain_local(self, X, y=None, name=None, init_score=None):
    """Provide local explanations for provided samples.
    Args:
        X: NumPy array for X to explain.
        y: NumPy vector for y to explain.
        name: User-defined explanation name.
        init_score: Optional. Either a model or per-sample initialization score.
    Returns:
        An explanation object, visualizing feature-value pairs
        for each sample as horizontal bar charts.
    """

Import

from interpret.glassbox import ExplainableBoostingClassifier
# Then: ebm.explain_local(X_test, y_test)

I/O Contract

Inputs

Name	Type	Required	Description
`X`	np.ndarray	Yes	NumPy array of samples to explain
`y`	np.ndarray	No	NumPy vector of true labels for the samples
`name`	str	No	User-defined explanation name
`init_score`	model / array / None	No	Either a model or per-sample initialization score

Outputs

Name	Type	Description
EBMExplanation	EBMExplanation	Per-sample feature contributions as horizontal bar charts

Usage Examples

Explaining Specific Samples

from interpret.glassbox import ExplainableBoostingClassifier
from interpret import show

ebm = ExplainableBoostingClassifier()
ebm.fit(X_train, y_train)

# Explain specific samples
local_exp = ebm.explain_local(X_test[:5], y_test[:5], name="Local Explanations")

# Visualize first sample's explanation
show(local_exp, key=0)

Related Pages

Page Connections

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

Principle

Implementation

Heuristic

Environment