Question 38
Domain 2: Evaluation, Tuning, and Quality OptimizationWhat parameter tuning strategy is most effective?
Correct answer: A
Explanation
Grid search is effective because it systematically evaluates every combination in the parameter ranges, so the best setting is chosen by the validation metric rather than guesswork. The process then uses a separate holdout test set to confirm performance and avoid overfitting to the validation data.
Why each option is right or wrong
A. Grid search over parameter ranges, evaluate each combination on validation set, select parameters maximizing target metric, validate on holdout test set.
Exhaustive grid search is the standard tuning procedure when the parameter space is finite and reasonably small: every candidate combination in the specified ranges is scored on the validation set, and the setting with the highest validation metric is retained. The final holdout test set is then used once, after selection, to estimate generalization performance without contaminating the tuning process; no specific statute or regulation applies here because this is a machine-learning methodology question rather than a legal one.
B. Manually try different parameters until results look good.
C. Use default parameters.
D. Tune one parameter at a time.