Several search candidates return NaN scores. How do you reveal the actual fit error instead of silently ranking them?
Classic MLLevel: IntermediateCandidate fitting failures
Several search candidates return NaN scores. How do you reveal the actual fit error instead of silently ranking them?
Hints
- NaN can mean failed fitting, not poor predictive quality.
- Turn failure recording into an exception.
Answer
Reproduce with error_score='raise' to surface the underlying exception. Check invalid parameter combinations, data requirements and memory limits before excluding candidates.
Source and licence: scikit-learn 1.7.2 — grid search · BSD-3-Clause · answer: Quantoros editorial