Why can increasing niter refine a continuous parameter search better than repeatedly drawing from three fixed values?
Classic MLLevel: IntermediateContinuous parameter coverage
Why can increasing n_iter refine a continuous parameter search better than repeatedly drawing from three fixed values?
Hints
- Sampling cannot invent values outside support.
- More trials and finer resolution are different.
Answer
A continuous distribution can propose intermediate values previously untested. A fixed list cannot explore beyond its support, no matter how large the draw budget becomes.
Source and licence: scikit-learn 1.7.2 — grid search · BSD-3-Clause · answer: Quantoros editorial