LIME Library Reference

LIME explains the predictions of any classifier in an interpretable and faithful manner by learning an interpretable model locally around the prediction.

Quick API (Tabular Data)

The core object is the LimeTabularExplainer. You must initialize it with the training data.

import lime.lime_tabular

explainer = lime.lime_tabular.LimeTabularExplainer(
    training_data=np_array_of_training_data,
    feature_names=['age', 'income', 'credit_score'],
    class_names=['Deny', 'Approve'],
    mode='classification' # or 'regression'
)

Extracting an Explanation

You explain a single instance (row) by passing the data and the model's prediction function.

exp = explainer.explain_instance(
    data_row=single_np_array_row, 
    predict_fn=model.predict_proba, # Must output probabilities for classification!
    num_features=5 # How many features to show in the explanation
)

# Render in a Jupyter notebook
exp.show_in_notebook(show_table=True)

# Or print as a list
print(exp.as_list())

KSB Mapping

KSB	Description	How This Addresses It
S5	Deployment, value assessment, and ROI	Translating model performance into business impact
S6	Communicate through storytelling and visualisation	Presenting ML results to non-technical stakeholders
B4	Consideration of organisational goals	Framing technical results in terms of business objectives
B1	Inquisitive approach	Exploring creative ways to explain model behaviour