Extra Trees
Train Extra Trees to predict categorical outcomes
Similar to Random Forest but uses random splits instead of optimal splits.
When to use:
- Want faster training than Random Forest
- Have large dataset
- Bias-variance tradeoff favors more randomness
Strengths: Faster training, even less overfitting than RF Weaknesses: Slightly less accurate than RF, still large models
Model Parameters
Same as Random Forest: n_estimators, max_depth, min_samples_split, min_samples_leaf, etc.