MLS-C01 Question 323
Select 3An e-commerce company wants to build a recommendation system to suggest products to its users based on their browsing and purchase history. They have a large labeled dataset of user activity and product interactions. The team is considering whether to use an Amazon SageMaker built-in algorithm or build a custom model. Which of the following would justify using a SageMaker built-in algorithm instead of building a custom model?
- A
The team wants to quickly prototype the solution without spending extensive time on model development.
- B
The dataset is highly unique, and the relationships between features require custom feature engineering and modeling.
- C
The business requires explainability for the model predictions, which can only be achieved with a custom-built model.
- D
The team has limited expertise in machine learning and wants to leverage pre-optimized algorithms for common machine learning tasks.
- E
The team is working on a tight deadline and needs a scalable solution with minimal setup and maintenance overhead.
Show answer and explanation
Correct answers: A, D, E
Explanation
SageMaker's built-in algorithms are pre-optimized for common machine learning tasks, reducing the time and expertise required to develop solutions. They are ideal for scenarios where rapid prototyping, scalability, and ease of use are priorities. On the other hand, custom models are more suitable when the problem requires unique feature engineering, domain-specific modeling, or advanced customization.
- A. Correct.
This is correct because SageMaker's built-in algorithms are pre-optimized and designed for quick prototyping, which reduces the time required to build a solution.
- B. Incorrect.
This is incorrect because highly unique datasets and complex relationships typically require custom models to effectively address the problem.
- C. Incorrect.
This is incorrect because explainability is not exclusive to custom-built models. Many SageMaker built-in algorithms also support explainability features (e.g., SHAP values for XGBoost).
- D. Correct.
This is correct because SageMaker built-in algorithms are designed for users with varying levels of machine learning expertise, making them ideal for teams with limited experience.
- E. Correct.
This is correct because SageMaker built-in algorithms provide scalable, production-ready solutions with minimal setup and maintenance, which is beneficial when working on tight deadlines.