Databricks Machine Learning Associate Question 323
Single answerYou are using hyperparameter tuning with a Random Forest model on Databricks to optimize its performance. After running 10 trials, you find that the model's accuracy has not improved significantly from the initial trials. What should you do next to potentially improve the model's accuracy?
- A
Increase the number of trials to explore a larger hyperparameter space.
- B
Reduce the number of trials to focus on fewer hyperparameter combinations.
- C
Manually select hyperparameters based on domain knowledge instead of running more trials.
- D
Increase the dataset size to provide the model with more training data.
Show answer and explanation
Correct answer: A
Explanation
The relationship between the number of trials and model accuracy in hyperparameter tuning is that more trials generally allow for a broader exploration of the hyperparameter space. This increases the likelihood of discovering hyperparameter combinations that improve the model's performance. In this scenario, increasing the number of trials is the best approach to potentially improve accuracy, especially if the current trials have not yielded significant improvements.
- A. Correct.
Increasing the number of trials allows the hyperparameter tuning process to explore a broader range of combinations and potentially find better-performing hyperparameters.
- B. Incorrect.
Reducing the number of trials decreases the likelihood of finding the optimal hyperparameters, as fewer combinations will be explored.
- C. Incorrect.
While domain knowledge can help in selecting hyperparameters, relying solely on manual selection instead of systematic tuning can overlook important combinations.
- D. Incorrect.
Increasing the dataset size can improve model performance in some situations, but it is unrelated to the number of trials during hyperparameter tuning.