A sparse count matrix fits in RAM until centering. Why, and how would you scale it safely?
Classic MLLevel: IntermediateSparse scaling
A sparse count matrix fits in RAM until centering. Why, and how would you scale it safely?
Hints
- Follow what happens to a zero entry.
- Separate centering from rescaling.
Answer
Subtracting nonzero means turns implicit zeros into nonzero entries. Use StandardScaler(with_mean=False) or MaxAbsScaler and retain sparse storage.
Source and licence: scikit-learn 1.7.2 — preprocessing · BSD-3-Clause · answer: Quantoros editorial