Load credit card transaction data and understand the extreme class imbalance inherent in fraud detection.
Apply techniques to handle the extreme class imbalance in fraud data — undersampling, SMOTE, and cost-sensitive learning.
Apply unsupervised anomaly detection methods — Isolation Forest, Local Outlier Factor, and One-Class SVM — to detect fraudulent transactions.
Compare supervised and unsupervised fraud detectors using precision-recall curves and business-relevant metrics.