MLA-C01 Question 128
Select 3You are building a machine learning pipeline to process large datasets and train a classification model using Amazon SageMaker. The dataset contains missing values in some columns, and you want to ensure the model performs well. Which of the following steps should you include in your pipeline to handle the missing data and optimize model performance?
- A
Use SageMaker Data Wrangler to impute missing values with the mean or median of the column
- B
Drop rows with missing values to simplify the dataset
- C
Use Amazon SageMaker Feature Store to store and retrieve imputed features
- D
Log missing data patterns for later analysis using Amazon CloudWatch
- E
Configure SageMaker Processing jobs to perform data imputation before training
Show answer and explanation
Correct answers: A, C, E
Explanation
Handling missing data is critical for building robust machine learning models. Using tools like SageMaker Data Wrangler and Processing jobs ensures that missing values are addressed effectively. Additionally, leveraging SageMaker Feature Store allows you to store and reuse the imputed features, streamlining the pipeline. Dropping rows with missing values can lead to data loss, and logging patterns, while helpful for monitoring, does not resolve the missing data problem directly.
- A. Correct.
Correct: SageMaker Data Wrangler provides built-in transformations, including imputation, that are essential for handling missing values in a preprocessing pipeline.
- B. Incorrect.
Incorrect: Dropping rows with missing values may lead to significant loss of data, which could degrade model performance unless the missing data is negligible.
- C. Correct.
Correct: Amazon SageMaker Feature Store can store processed features, such as imputed values, so they can be reused across training and inference pipelines.
- D. Incorrect.
Incorrect: While logging missing data patterns can be useful for monitoring, it does not directly address the issue of handling missing data during preprocessing.
- E. Correct.
Correct: SageMaker Processing jobs can handle data preprocessing tasks, such as imputation, at scale, ensuring the dataset is clean before training begins.