A random forest considers a feature subset at its root split. Are all descendants restricted to that same initially…
Classic MLLevel: IntermediateFeature sampling granularity
A random forest considers a feature subset at its root split. Are all descendants restricted to that same initially chosen subset?
Hints
- Identify the stage at which max_features applies.
- Node-level and tree-level sampling are different designs.
Answer
No. Random feature selection occurs when splitting each node, rather than selecting one permanent feature set for the whole tree. Different nodes can therefore consider different candidate features.
Source and licence: scikit-learn 1.7.2 — ensemble · BSD-3-Clause · answer: Quantoros editorial