Why can fitting a tree to a large float64 array cause an additional memory allocation?
Classic MLLevel: IntermediateTree input memory
Why can fitting a tree to a large float64 array cause an additional memory allocation?
Hints
- Internal dtype may differ from input dtype.
- Peak memory includes conversion buffers.
Answer
The documented tree implementation converts inputs to float32 internally. Supplying a compatible representation may avoid a conversion copy, but confirm precision and sparse-format requirements.
Source and licence: scikit-learn 1.7.2 — tree · BSD-3-Clause · answer: Quantoros editorial