MLS-C01 Question 322
Single answerA data science team is tasked with building a machine learning model to predict customer churn for a subscription-based service. The team has access to structured tabular data with a mix of numerical and categorical features. They also have limited expertise in deep learning but are experienced in using AWS services. Which of the following approaches is the most appropriate for building the model?
- A
Use Amazon SageMaker's built-in XGBoost algorithm to train the model on the structured data.
- B
Build a custom deep learning model using TensorFlow and deploy it using Amazon SageMaker.
- C
Use Amazon SageMaker's built-in object detection algorithm to analyze customer behavioral patterns.
- D
Develop a custom reinforcement learning model using Amazon SageMaker RL and train it to predict churn.
Show answer and explanation
Correct answer: A
Explanation
When working with structured tabular data, the XGBoost algorithm is a strong choice due to its ability to handle numerical and categorical features effectively. SageMaker's built-in algorithms, like XGBoost, allow data science teams with limited expertise in deep learning to quickly build and deploy accurate machine learning models without the need for custom development. Other approaches, such as deep learning or reinforcement learning, are either unnecessary or unsuitable for this scenario.
- A. Correct.
Amazon SageMaker's built-in XGBoost algorithm is well-suited for structured data, such as tabular data with numerical and categorical features. It is a scalable, efficient, and proven solution for predictive tasks like churn prediction.
- B. Incorrect.
Building a custom deep learning model requires expertise in deep learning frameworks like TensorFlow and might not provide better performance for structured tabular data compared to algorithms like XGBoost.
- C. Incorrect.
The object detection algorithm is designed for image data and is not suitable for structured tabular data or churn prediction tasks.
- D. Incorrect.
Reinforcement learning is typically used for sequential decision-making problems, not for predictive tasks like churn prediction.