Free Databricks Machine Learning Associate Actual Exam Questions - Question 10 Discussion

C for sure, it’s the only one that truly adapts based on past evaluations.

C That method builds a model of the objective function and uses it to pick promising hyperparameters, unlike random or grid search which don’t learn from previous results.

B tbh feels off since it mainly just cuts down the search space randomly without really learning from past results. C seems spot on because it actually builds a probabilistic model based on previous trials and uses that to pick the next promising hyperparameters. A and D just don’t adapt—they either pick randomly or exhaustively. So, it really boils down to C being the only one actively using past data to guide future sampling.

Maybe D could be ruled out quickly since Grid Search just tries every combination without learning from past trials. Random Search (A) is also random, so no learning there. Halving Random Search (B) does some pruning but still doesn’t really use previous trial info to pick new hyperparameters. So yeah, C fits best because it’s the only one that adapts based on what it’s already tried.

C imo, because it builds probabilistic models to guide the search efficiently.

C - Does it rely on Bayesian optimization techniques?