Databricks Machine Learning Associate Question 233
Select 3A data scientist is working on a machine learning model and is deciding between using cross-validation or a simple train-validation split for model evaluation. Which of the following are valid considerations when using cross-validation instead of a train-validation split?
- A
Cross-validation provides a more reliable estimate of model performance by averaging results over multiple splits.
- B
Cross-validation requires less computational resources compared to a train-validation split.
- C
Cross-validation can reduce the risk of selecting a model that performs well only on a specific data split.
- D
Cross-validation is faster to execute when working with large datasets.
- E
Cross-validation may take significantly more time to compute compared to a train-validation split.
Show answer and explanation
Correct answers: A, C, E
Explanation
Cross-validation is a technique that provides a more robust evaluation of a model by using multiple splits of the data, which helps mitigate overfitting to a specific split. However, this comes at the cost of higher computational requirements and longer execution times compared to a simple train-validation split. These trade-offs should be considered when selecting an evaluation method.
- A. Correct.
This is correct. Cross-validation evaluates the model on multiple splits of the data, leading to a more robust estimate of model performance.
- B. Incorrect.
This is incorrect. Cross-validation generally requires more computational resources because the model is trained and validated multiple times, unlike a single train-validation split.
- C. Correct.
This is correct. By using multiple data splits, cross-validation reduces the risk of overfitting to a specific split of the data.
- D. Incorrect.
This is incorrect. Cross-validation is computationally intensive, especially with large datasets, as it involves multiple training and validation cycles.
- E. Correct.
This is correct. Cross-validation requires repeated training and validation, which increases computation time compared to a single train-validation split.