Why does one-vs-rest multiclass calibration need a final probability normalization step?
Classic MLLevel: AdvancedNormalizing calibrated classes
Why does one-vs-rest multiclass calibration need a final probability normalization step?
Hints
- Independent binary probabilities have no shared simplex constraint.
- Prediction requires a coherent class distribution.
Answer
This answer is part of mentoring
Advanced questions come with a full breakdown and a mock interview with a working quant researcher. The first call is free.
Source and licence: scikit-learn 1.7.2 — calibration · BSD-3-Clause · answer: Quantoros editorial