lime_ndt

lime_ndt is a Python library that introduces an enhanced version of the LIME technique for model explainability, leveraging Neural Decision Trees (NDTs) for improved local and global interpretability of machine learning models.

Installation

pip install lime_ndt

Quick Start (Classification)

from sklearn.datasets import load_iris
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split
from lime_ndt import LimeNdtExplainer

# Load data and train a model
iris = load_iris()
X_train, X_test, y_train, y_test = train_test_split(iris.data, iris.target, random_state=42)
rf = RandomForestClassifier(random_state=42).fit(X_train, y_train)

# Explain a prediction (NDT surrogate is used by default)
explainer = LimeNdtExplainer(X_train, feature_names=iris.feature_names,
                              class_names=iris.target_names.tolist(), mode='classification')
exp = explainer.explain_instance(X_test[0], rf.predict_proba)

# Visualize
exp.as_pyplot_figure()

Regression Example

from sklearn.datasets import load_diabetes
from sklearn.ensemble import RandomForestRegressor
from sklearn.model_selection import train_test_split
from lime_ndt import LimeNdtExplainer

# Load data and train a model
diabetes = load_diabetes()
X_train, X_test, y_train, y_test = train_test_split(diabetes.data, diabetes.target, random_state=42)
rf = RandomForestRegressor(random_state=42).fit(X_train, y_train)

# Explain a prediction
explainer = LimeNdtExplainer(X_train, feature_names=diabetes.feature_names,
                              discretize_continuous=True, mode='regression')
exp = explainer.explain_instance(X_test[0], rf.predict, num_features=10)

# Show results
print(exp.as_list())
exp.as_pyplot_figure()

Advanced Usage

You can customize the NDT surrogate parameters (gammas, tree depth, training epochs):

from sklearn.datasets import load_iris
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split
from lime_ndt import LimeNdtExplainer, NDTRegressorWrapper

# Load data and train a model
iris = load_iris()
X_train, X_test, y_train, y_test = train_test_split(iris.data, iris.target, random_state=42)
rf = RandomForestClassifier(random_state=42).fit(X_train, y_train)

# Create explainer
explainer = LimeNdtExplainer(X_train, feature_names=iris.feature_names,
                              class_names=iris.target_names.tolist(), mode='classification')

# Custom NDT surrogate with specific parameters
ndt = NDTRegressorWrapper(gammas=[1, 100], max_depth=10, epochs=50)
exp = explainer.explain_instance(X_test[0], rf.predict_proba, model_regressor=ndt)

print(exp.as_list())

Features

• Enhanced LIME explanations using Neural Decision Trees.
• Support for tabular data (classification and regression).
• NDT surrogate used by default — no extra setup needed.
• Tools for extracting, analyzing, and comparing decision trees and forests.
• Utilities for discretization (quartile, decile, entropy).
• Integration with Keras and scikit-learn.

Requirements

• Python 3.9+
• numpy, pandas, scikit-learn, keras, tensorflow, matplotlib

How to cite

Mohamed Aymen Bouyahia and Argyris Kalogeratos, Enhancing LIME using Neural Decision Trees, arXiv preprint, http://arxiv.org/abs/2603.20919, 2026.

License

This project is licensed under the Apache License 2.0. See the LICENSE file for details.