Skip to main content
Random Forest trains many decision trees on random subsets of your data and combines their votes into a final prediction. It handles mixed feature types well, is resistant to overfitting, and requires minimal tuning.

​
Configuration

ParameterDescriptionDefault
N EstimatorsNumber of trees in the forest. More trees = more stable but slower.100
Max DepthMaximum depth of each tree. Leave empty for unlimited.None
Min Samples SplitMinimum samples required to split a node.2
Min Samples LeafMinimum samples required in a leaf node.1
Random StateSeed for reproducibility.42

​
Input / Output

Type
InputX Train + Y Train (from Train-Test Split)
OutputTrained Model

​
When to use

A good default choice for most classification problems. Works well on tabular data without feature scaling.