FraudVision

This project aims to develop a machine learning model to detect fraudulent credit card transactions in real-time. The dataset contains 31 features and a class label. These features include "Time" (seconds elapsed between transactions), 28 anonymized PCA-transformed features (V1 to V28), "Amount" (transaction amount in USD), and a "Class" label indicating whether a transaction is fraudulent (1) or not (0). This project leverages the capabilities of Streamlit, a powerful open-source app framework, to build an interactive and intuitive web application that enables users to detect and analyze fraudulent credit card transactions in real-time. A Logistic Regression model was trained on the dataset. The model achieved a training accuracy of 93% and a testing accuracy of 89%. The project successfully developed a high-accuracy fraud detection model using Logistic Regression. Future work will focus on enhancing the model with additional features and integrating it into real-world financial systems for live fraud detection, thus significantly contributing to the security of credit card transactions..