How does SGDClassifier extend a binary linear loss to several classes?
Classic MLLevel: IntermediateSGD multiclass construction
How does SGDClassifier extend a binary linear loss to several classes?
Hints
- Each class is contrasted with all others.
- Check output-to-class ordering.
Answer
It fits one-versus-rest binary models and compares their decision scores. The coefficient matrix has one row per class in the fitted classes_ order.
Source and licence: scikit-learn 1.7.2 — sgd · BSD-3-Clause · answer: Quantoros editorial