You are training an ML model using data stored in BigQuery that contains several values that are considered Personally Identifiable Information (PII). You need to reduce the sensitivity of the…

Question

Options

• AUsing Dataflow, ingest the columns with sensitive data from BigQuery, and then randomize the
• BUse the Cloud Data Loss Prevention (DLP) API to scan for sensitive data, and use Dataflow with
• CUse the Cloud Data Loss Prevention (DLP) API to scan for sensitive data, and use Dataflow to
• DBefore training, use BigQuery to select only the columns that do not contain sensitive data.

• A
  11% (2)
• B
  83% (15)
• D
  6% (1)

Explanation

The Cloud Data Loss Prevention (DLP) API is Google's purpose-built tool for discovering and classifying sensitive data. After identifying PII fields, Dataflow applies de-identification transformations - such as tokenization, pseudonymization, or masking - that replace sensitive values with non-sensitive surrogates while preserving the column and its statistical utility for model training. This satisfies the constraint that 'every column is critical.' Option C implies removing data (not just transforming it), which would eliminate columns. Option D explicitly drops sensitive columns, violating the constraint. Option A uses randomization, which destroys the statistical relationships the model needs to learn.

No community discussion yet for this question.