Build and evaluate classification models — class balance, metrics (precision/recall/ROC), thresholds, and error costs. Use when predicting categories such as churn, fraud, conversion, or risk tiers.
