MLA-C01 Question 370
Select 3You are developing a machine learning model to predict customer churn for a subscription-based service using Amazon SageMaker. The dataset contains missing values and categorical features. Which of the following steps should you take to prepare the data for training?
- A
Use Amazon SageMaker Data Wrangler to impute missing values and encode categorical features.
- B
Drop rows with missing values to ensure clean data is used for training.
- C
Use one-hot encoding to transform categorical features into numerical representations.
- D
Normalize the numerical features to ensure all features are on the same scale.
- E
Utilize SageMaker's built-in algorithms, which automatically handle missing data and categorical features without preprocessing.
Show answer and explanation
Correct answers: A, C, D
Explanation
To prepare the dataset for training, it is essential to handle missing values, transform categorical features, and normalize numerical features. Amazon SageMaker Data Wrangler can simplify these preprocessing tasks. Dropping rows with missing values is not ideal as it leads to loss of potentially valuable data, and relying solely on built-in algorithms for preprocessing may not cover all edge cases. Proper preprocessing ensures the model can effectively learn patterns in the data.
- A. Correct.
Amazon SageMaker Data Wrangler provides tools to impute missing values and encode categorical features, making this an appropriate step for data preprocessing.
- B. Incorrect.
Dropping rows with missing values can result in significant data loss, which may negatively impact model performance. It is generally not the best practice unless missing data is minimal.
- C. Correct.
One-hot encoding is a standard method to transform categorical features into numerical values, which is required for most machine learning algorithms.
- D. Correct.
Normalizing numerical features ensures that all features have similar scales, which is especially important for algorithms sensitive to feature magnitude, such as gradient-based methods.
- E. Incorrect.
While SageMaker's built-in algorithms can handle some preprocessing, it is not guaranteed they will address all missing value and categorical feature issues. Explicit preprocessing is often necessary for optimal performance.