GCP Generative AI Leader Practice Question 43: Data Consistency Across Sources
Google Cloud Generative AI Leader · Test 5 · question 3 of 10
A logistics company wants to train a model on shipment records pulled from three regional warehouse systems. One system logs weight in kilograms, another in pounds. One system records delivery status as "Delivered" or "Pending", another uses numeric codes 1 and 0 for the same two outcomes. Every record is present and every field is filled in.
Which data quality problem does this combined dataset have, and what does it threaten?
Show answer and explanation
Answer:
Consistency means the same fact is represented the same way everywhere it appears. Here, weight is recorded in two different units and delivery status is recorded with two different labeling schemes for the identical two outcomes. A model trained directly on this mix will treat "Delivered" and the code "1" as unrelated categories and will average weights that are not on the same scale, so the data must be standardized (common units, common labels) before training.
Option B is wrong because the scenario states every field is filled in; nothing is missing, so completeness is not the problem. Option C is wrong because the scenario never mentions the age of the records, only how they are formatted. Option D is wrong because the records are described as available and readable, just inconsistently formatted; accessibility problems are about not being able to reach the data at all, which is not the case here.
Reference: Auto data quality overview