RFGBoost

RFGBoost combines Random Forest with Gradient Boosting using Weight of Evidence encoding for interpretable machine learning with categorical features.

Key Features

• Interpretable Boosting: Uses Random Forest as base learners instead of single decision trees
• Categorical Feature Mastery: Built-in WOE encoding handles high-cardinality categories automatically
• Uncertainty Quantification: Built-in confidence intervals for risk-aware predictions
• Dual Base Learners: Choose between scikit-learn or XGBoost RandomForest implementations

Best Used For

• Financial Risk Modeling: Credit scoring and risk management where interpretability is mandatory
• Categorical-Heavy Datasets: Datasets with many categorical features that need careful encoding
• Regulatory Compliance: Applications requiring clear decision paths and explainable predictions

Perfect for data scientists who need gradient boosting performance with the interpretability that Random Forest provides, especially in regulated industries.