Credit Card Fraud Detection System Using Django and Machine Learning

Abstract

The Credit Card Fraud Detection System is designed to identify and prevent fraudulent credit card transactions using Machine Learning algorithms integrated with the Django framework. The project detects unusual transaction behaviour based on parameters such as transaction amount, merchant category, and transaction country. Fraudulent activities have become increasingly complex, and traditional detection systems often fail to identify evolving fraud patterns; this work therefore leverages data analytics and machine learning to improve the accuracy and speed of fraud detection. The system uses classification algorithms such as Logistic Regression, Decision Tree, and Random Forest, trained on a dataset containing both legitimate and fraudulent transactions, to predict the probability of a transaction being fraudulent in real time, with transactions exceeding a configured probability threshold flagged as suspicious. A Django web interface allows users and administrators to view flagged transactions, train models, and test new data, with role-based access control that supports three roles: Administrator, Analyst, and Reviewer, each with specific permissions. The trained model is serialized using Joblib or Pickle and integrated with Django for real-time prediction, and the application is structured in a presentation– application–data layered architecture. By combining machine learning with web-based delivery and role-based security, the project demonstrates a secure, efficient, and userfriendly platform that reduces false positives, improves detection accuracy, and automates the monitoring process while ensuring data privacy and scalability for real-world use.