PROFESSIONAL-MACHINE-LEARNING-ENGINEER · Question #150
PROFESSIONAL-MACHINE-LEARNING-ENGINEER Question #150: Real Exam Question with Answer & Explanation
The correct answer is B: Develop a simple heuristic (e.g., based on z-score) to label the machines' historical performance. Before any ML model can be trained for predictive maintenance, the historical data must be labeled (i.e., incidents must be marked as failures or anomalies). Since hiring analysts (D) is expensive and slow, and training a supervised model without labels (A) is impossible, the fas
Question
You work on a team in a data center that is responsible for server maintenance. Your management team wants you to build a predictive maintenance solution that uses monitoring data to detect potential server failures. Incident data has not been labeled yet. What should you do first?
Options
- ATrain a time-series model to predict the machines' performance values. Configure an alert if a
- BDevelop a simple heuristic (e.g., based on z-score) to label the machines' historical performance
- CDevelop a simple heuristic (e.g., based on z-score) to label the machines' historical performance
- DHire a team of qualified analysts to review and label the machines' historical performance data.
Explanation
Before any ML model can be trained for predictive maintenance, the historical data must be labeled (i.e., incidents must be marked as failures or anomalies). Since hiring analysts (D) is expensive and slow, and training a supervised model without labels (A) is impossible, the fastest and cheapest first step is to create approximate labels using a simple heuristic such as a z-score threshold on performance metrics. These heuristic labels bootstrap the labeling process and allow initial model training, which can later be refined with higher-quality labels. This approach balances speed, cost, and practicality.
Topics
Community Discussion
No community discussion yet for this question.