Handling an Imbalanced Dataset

Question: How would you handle an imbalanced dataset? Key points: Techniques include resampling (oversampling the minority class e.g. SMOTE, or undersampling the majority class), class-weighted loss functions to penalize minority-class errors more heavily, using threshold tuning instead of default 0.5, and choosing evaluation metrics robust to imbalance (F1, precision/recall, PR-AUC) rather than raw accuracy. Ensemble techniques (e.g. balanced bagging) can also help.