MLA-C01 Question 248
Select 3You are tasked with building a machine learning model to predict customer churn for a subscription-based service. The dataset contains millions of records and is stored in Amazon S3 in CSV format. You need to preprocess the data and train a model using Amazon SageMaker while ensuring that the process is efficient. Which of the following strategies should you adopt to meet these requirements?
- A
Use Amazon SageMaker Processing jobs to preprocess the data directly from Amazon S3.
- B
Load the entire dataset into a Jupyter Notebook instance before preprocessing it locally.
- C
Use Amazon SageMaker Pipe mode to stream data directly from Amazon S3 during training.
- D
Convert the dataset into Apache Parquet format to optimize I/O and storage performance.
- E
Manually split the dataset into smaller chunks and upload them to SageMaker for processing.
Show answer and explanation
Correct answers: A, C, D
Explanation
To efficiently preprocess and train a machine learning model on a large dataset stored in Amazon S3, it is essential to leverage scalable and optimized solutions provided by AWS. SageMaker Processing jobs simplify preprocessing tasks by operating directly on data in S3. Using Apache Parquet format reduces storage and I/O overhead. During training, SageMaker Pipe mode streams data directly from S3, further improving efficiency. Avoid local preprocessing or manual dataset splitting as these approaches are less scalable and prone to errors.
- A. Correct.
Amazon SageMaker Processing jobs allow you to preprocess data directly from Amazon S3 in a distributed and scalable manner, making it an efficient choice for large datasets.
- B. Incorrect.
Loading the entire dataset into a Jupyter Notebook instance for local preprocessing is inefficient for large datasets and may lead to memory issues.
- C. Correct.
Amazon SageMaker Pipe mode streams data directly from S3 during training, which reduces the need for downloading the entire dataset and optimizes training performance.
- D. Correct.
Converting the dataset into Apache Parquet format reduces storage costs and improves I/O performance, making it a suitable preprocessing step for large-scale datasets.
- E. Incorrect.
Manually splitting the dataset into smaller chunks is time-consuming and error-prone. SageMaker provides built-in capabilities for handling large datasets more efficiently.