MLA-C01 Practice Question 41: Glue Data Quality Rules
AWS Certified Machine Learning Engineer – Associate (MLA-C01) · Test 5 · question 1 of 10
An AWS Glue ETL job builds the daily training dataset for a demand-forecasting model. Twice last quarter an upstream system shipped a bad extract, once with unit_price empty for half the rows and once with duplicate order_id values. The job loaded both extracts, and the retrained model degraded before anyone noticed. The team wants the pipeline itself to check each day's data against explicit rules and keep failing records out of the training dataset, without running a separate service or standing up infrastructure.
What should the team do?
Show answer and explanation
Answer:
AWS Glue Data Quality is serverless and runs inside Glue ETL jobs. You write rules in the Data Quality Definition Language (DQDL), for example IsComplete "unit_price" and IsUnique "order_id", and the job evaluates them as data flows through. In ETL jobs it can identify the exact records that failed, so the job can quarantine them before they reach the training dataset. That catches both past incidents inside the pipeline, with no extra infrastructure (option C).
- A is wrong because Model Monitor watches data sent to a deployed endpoint for inference. It does not check the training dataset, so a bad extract would still reach retraining. - B is wrong because both bad extracts could have a normal row count. Empty prices and duplicate IDs are problems with values, which a record count cannot see, and an alarm only notifies someone without filtering anything. - D is wrong because a crawler tracks schema, meaning columns and types. An empty unit_price column and duplicate order_id values leave the schema unchanged.
Reference: AWS Glue Data Quality