Random Forest
Train diverse trees, inspect their votes, and see the aggregate decision stabilize.
Random Forest reduces dependence on one tree by combining many randomized decision trees. ModelOrigen shows actual tree structures, sampling, individual predictions, votes, and aggregate metrics.
Step by step
- Create a train/test split.
- Sample rows (bootstrap when enabled).
- Sample candidate features for a tree.
- Grow a real decision tree.
- Repeat for many trees.
- Vote for classification or average for regression.
Core formulas
Classification aggregate
ŷ = mode(T₁(x), …, Tᴮ(x))The most common tree vote becomes the class.
Regression aggregate
ŷ = (1/B) Σ T_b(x)Tree predictions are averaged for regression.
When to use Random Forest
- Nonlinear interactions are likely.
- A robust tree ensemble is more important than one fully interpretable tree.
- You want a strong general-purpose baseline with minimal scaling concerns.