Which observations can evaluate a particular bootstrap tree without being in that tree's training sample?
Classic MLLevel: IntermediateOut-of-bag evidence
Which observations can evaluate a particular bootstrap tree without being in that tree's training sample?
Hints
- Bootstrap sampling leaves some rows out.
- Exclusion must be checked per tree and row.
Answer
Its out-of-bag observations, absent from that bootstrap draw. Aggregate only predictions from trees that excluded each row; OOB is not automatically valid for dependent groups or temporal deployment.
Source and licence: scikit-learn 1.7.2 — ensemble · BSD-3-Clause · answer: Quantoros editorial