Fundamentals of AI and ML
ML Lifecycle
The ML lifecycle runs from business framing through monitoring. Deployment is not the end: drift, quality, latency, and business outcomes need ongoing review.
Concepts
- Typical steps include business framing, data collection, data preparation, feature engineering, training, evaluation, deployment, monitoring, and iteration.
- Feature engineering transforms raw data into useful model inputs.
- Monitoring watches for data drift, model drift, quality regressions, latency, and operational issues.
Exam tips
- Feature engineering turns raw data into useful model inputs.
- Monitoring can detect data drift, model drift, quality changes, and operational issues.
- Iteration is normal when data, requirements, or model behavior changes.