subtopic

Classification: class balance (Project A)

Plot class counts and percentages; state the imbalance ratio and what it means for metrics and splits. Concepts: Class imbalance; stratification; baseline accuracy Project A: 700 Good / 300 Bad (30% bad, ratio 2.33:1) → moderate imbalance. A model that always says 'Good' scores 70% accuracy and catches zero Bad applicants. Plan: stratified split, class_weight, cost-based threshold (E13). Project B: Not applicable (continuous target). Equivalent idea: check how much of the error will come from rare peak hours.