Black Box PrecisionCore SDK

API Reference

Complete reference for the BlackBox Precision Core SDK API methods and classes.

Explainer

The main class for generating explanations for black-box models.

Constructor

python
Explainer(model, method='shap', background_data=None, **kwargs)
Parameters:
  • model - Your trained ML model (required)
  • method - Explanation method: 'shap' or 'lime' (default: 'shap')
  • background_data - Background dataset for SHAP (optional)

explain()

Generate explanations for input data.

python
explain(data, feature_names=None, num_features=10)
Parameters:
  • data - Input data to explain (numpy array or pandas DataFrame)
  • feature_names - List of feature names (optional)
  • num_features - Number of top features to show (default: 10)
Returns:

Explanation object with feature importances and visualizations

global_explanation()

Generate global feature importance across the entire dataset.

python
global_explanation(data, feature_names=None)

Explanation Object

Object returned by explain() containing explanation results and visualization methods.

Attributes

  • values

    Feature importance values (numpy array)

  • feature_names

    List of feature names

  • base_value

    Model's base prediction value

Methods

  • plot()

    Visualize feature importances

  • to_dict()

    Export explanation as dictionary

  • save(filename)

    Save explanation to file

Complete Example

python
from blackbox_core import Explainer
import numpy as np
from sklearn.ensemble import RandomForestClassifier

# Train a model
X_train = np.random.rand(100, 10)
y_train = np.random.randint(0, 2, 100)
model = RandomForestClassifier()
model.fit(X_train, y_train)

# Initialize explainer
explainer = Explainer(model, method='shap')

# Get explanation
X_test = np.random.rand(1, 10)
explanation = explainer.explain(
    X_test,
    feature_names=[f'feature_{i}' for i in range(10)]
)

# Visualize
explanation.plot()

# Export
results = explanation.to_dict()