CardSheild: A Credit Card Fraud Detection System
Sayala Guru Preethika -, Damala Sushma Sri -, MD Aman Ahmed -, Ponnuru Meghana · International Journal on Science and Technology · 2025
This project presents a machine learning-based credit card fraud detection system using Supervised algorithms (Logistic Regression, Decision Trees, Random Forest, SVM, XGBoost, ANN) and Unsupervised techniques (Autoencoders, Isolation Forests) for anomaly detection. To address data imbalance, SMOTE, oversampling, and undersampling are applied. The workflow includes preprocessing, EDA, model training, and evaluation using Accuracy, Precision, Recall, and AUC-ROC, with hyperparameter tuning for optimization. The best model is deployed via a web app/API for real-time detection, integrating with banking systems. The system enhances accuracy, reduces false positives, and supports transparency with explainable AI, offering a scalable, secure solution for financial fraud prevention.