MLA-C01 Question 109
Select 3A healthcare company is building a machine learning model to analyze patient data and predict the likelihood of certain diseases. To comply with privacy regulations, the company must ensure that personally identifiable information (PII) is removed or obfuscated before the data is processed. Which of the following techniques would be most appropriate for this purpose?
- A
Use AWS Glue DataBrew to identify and mask PII fields in the dataset.
- B
Perform data anonymization by assigning unique identifiers to replace PII.
- C
Encrypt the entire dataset using AWS Key Management Service (KMS) before processing.
- D
Use Amazon Macie to detect and redact sensitive data automatically.
- E
Implement access control policies in Amazon S3 to limit access to the dataset.
Show answer and explanation
Correct answers: A, B, D
Explanation
To comply with privacy regulations, the dataset must have PII removed or obfuscated before processing. AWS Glue DataBrew can mask PII fields, anonymization replaces PII with non-identifiable data while retaining utility, and Amazon Macie can detect and redact sensitive information. Encryption and access control are important for data security but do not fulfill the specific requirement of removing or obfuscating PII.
- A. Correct.
Correct. AWS Glue DataBrew provides built-in transformations to identify and mask PII, making it a suitable tool for this use case.
- B. Correct.
Correct. Data anonymization, such as replacing PII with unique identifiers, ensures that sensitive information is not directly tied to individuals while maintaining data usability.
- C. Incorrect.
Incorrect. Encrypting the dataset protects it during storage and transmission but does not remove or obfuscate PII, which is required for compliance in this scenario.
- D. Correct.
Correct. Amazon Macie can detect and redact sensitive data, making it an effective tool for identifying and managing PII in datasets.
- E. Incorrect.
Incorrect. While access control policies enhance data security by restricting access, they do not address the requirement to remove or obfuscate PII in the dataset.