Skip to content
Quantoros

Which simple scaler keeps zero exactly zero without estimating a mean?

Classic MLLevel: FoundationZero-preserving scaling

Which simple scaler keeps zero exactly zero without estimating a mean?

Hints
  • Do not subtract a location parameter.
  • Divide by one magnitude per column.

Answer

MaxAbsScaler divides each feature by its maximum absolute training value. It preserves zeros and signs, but is sensitive to extreme values.

Source and licence: scikit-learn 1.7.2 — preprocessing · BSD-3-Clause · answer: Quantoros editorial

Related questions