Classification
A supervised learning technique in machine learning where the model predicts the category or class label of new observations based on training data.
Definition
Involves training on labeled examples (features → class) to learn decision boundaries. Classification can be binary (spam vs. not-spam) or multi-class (handwriting recognition). Governance practice includes validating class balance, monitoring per-class accuracy, and updating models as class definitions evolve (e.g., new fraud types).
Real-World Example
A bank’s fraud team deploys a classifier that flags transactions as “legitimate,” “fraud,” or “review.” They train it on historical labeled transactions, monitor false-positive rates per category, and retrain every quarter to capture emerging fraud patterns.
.svg)
