This project uses unsupervised clustering via DBSCAN to detect fraudulent transactions without any prior labeling. By leveraging Z-score standardization, PCA for visualization, and cluster evaluation metrics, we create an effective and interpretable anomaly detection pipeline.

End-to-end Credit Card Fraud Detection system using Supervised ML models and Deep Autoencoders for anomaly detection. Includes advanced imbalance handling, extensive visual analysis (PCA, t-SNE, reconstruction errors), and a fully interactive Streamlit app for real-time fraud prediction.