Which phase ensures models remain accurate over time?

• A. Deployment
• B. Evaluation
• C. Monitoring
• D. Data Preparation

Answer: (C)

Maintaining accuracy over time comes from continuous, in-production monitoring. After a model is deployed, the real world can change—the data it sees may drift, relationships can evolve, and input patterns can shift. Monitoring tracks ongoing performance metrics (like accuracy, precision, recall, AUC), data distribution changes, and system health. When signs of drift or degradation appear, you can trigger retraining, feature adjustments, or model replacement to keep performance up to date. Deployment, evaluation, and data preparation play important roles at different stages, but only monitoring focuses on sustaining accuracy once the model is in use.