Question 870
At an enterprise inventorying information for a new model, an AI project leader faces this situation: An energy consumption model receives building-level monthly totals but must optimize individual machines every minute. Available records are not automatically relevant or authorized for training. What must be checked so the correction does not create a new problem?
Explanation
Correct Answer
In an enterprise inventorying information for a new model, the key precaution is: Do not infer device-level behavior directly from aggregated building totals without validation. The planned action is: Agree asset-level sampling and interval requirements with engineers and source owners. This caution matters because: Prediction granularity must match the decision unit and timing the business wants to support. The team should still verify: A data sample has sufficient per-machine time resolution to support the intended recommendation.
Incorrect Answers
Cloud repositories require dataset catalog, ownership and lineage to distinguish raw from approved sources. The appropriate correction for the reported case is: Agree asset-level sampling and interval requirements with engineers and source owners.
Data collection must define eligibility windows and reconciliation for delayed events. Fixing that different mechanism would not address the current root cause: prediction granularity must match the decision unit and timing the business wants to support.
Data-provider communication and change ownership reduce unexpected schema and quality failures. Its verification is distinct from the reported condition: An energy consumption model receives building-level monthly totals but must optimize individual machines every minute. Available records are not automatically relevant or authorized for training.