Random Forest

Random Forest is an ensemble learning method for classification and regression with GPU acceleration support.

Classification

Train a Random Forest classifier:

Copy
CREATE TEMP TABLE rf_model AS
SELECT neurondb.train(
    'default',
    'random_forest',
    'training_table',
    'label',
    ARRAY['features'],
    '{"n_trees": 3}'::jsonb
)::integer AS model_id;

Regression

Train a Random Forest regressor:

Copy
CREATE TEMP TABLE rf_model AS
SELECT neurondb.train(
    'default',
    'random_forest',
    'training_table',
    'target',
    ARRAY['features'],
    '{"n_trees": 3}'::jsonb
)::integer AS model_id;

Prediction

Copy
SELECT neurondb.predict(
    (SELECT model_id FROM rf_model),
    features
) AS prediction
FROM test_table;

Learn More

For detailed documentation on Random Forest parameters, hyperparameter tuning, feature importance, and GPU optimization, visit:

Random Forest Documentation

Related Topics

• Classification - Other classification algorithms
• Model Management - Managing trained models