Stack Explorer

CatBoost

machine-learning

Yandex gradient boosting library optimized for categorical features

1.5M/week → Stable

Features

boostingcategoricalyandextabulargradient-boosting

Pros and Cons

Ventajas

  • + Best handling of categorical features
  • + Less tuning needed than XGBoost
  • + Robustness against overfitting
  • + Excellent GPU support
  • + Symmetric trees for better generalization
  • + Good defaults out of the box

Desventajas

  • - Slower than LightGBM in some cases
  • - Larger binaries
  • - Less popular than XGBoost
  • - Less extensive documentation
  • - Fewer integrations available

Use Cases

  • Data with many categorical features
  • Fast classification and regression
  • Problems where XGBoost overfits
  • Click-through rate prediction
  • Recommendation systems
  • Applications with mixed data