Why is calling partialfit once per individual document not always efficient for naive Bayes?
Classic MLLevel: IntermediateIncremental batch sizing
Why is calling partial_fit once per individual document not always efficient for naive Bayes?
Hints
- Calls have fixed costs besides arithmetic.
- Streaming need not mean single-row updates.
Answer
Each call has overhead. Larger chunks within the memory budget amortize that overhead while retaining incremental learning; batch size is a throughput–memory trade-off.
Source and licence: scikit-learn 1.7.2 — naive bayes · BSD-3-Clause · answer: Quantoros editorial