💳 Credit Card Fraud Detection System

Deep Learning–Based Anomaly Detection for Real-Time Transaction Security

An Autoencoder-powered fraud detection engine that flags suspicious online credit card transactions in real time, wrapped in an interactive Streamlit interface.

Live Demo

---

📌 Overview

A web-based fraud detection app built with Streamlit, TensorFlow, and Python, using an Autoencoder trained to learn normal transaction behavior. Transactions it reconstructs poorly — i.e. that deviate from learned normal patterns — are flagged as potential fraud.

This anomaly-detection framing sidesteps the class-imbalance problem that plagues standard fraud classifiers: rather than needing enough labeled fraud examples to learn from, the model just learns what normal looks like.

Features: real-time fraud prediction · transaction data visualization · interactive dataset exploration

---

Credit Card Fraud Detection System

Deep learning-based fraud detection system for identifying suspicious online credit card transactions in real time.

---

Overview

The Credit Card Fraud Detection System is a web-based machine learning application designed to analyze online financial transactions and detect potentially fraudulent activities.

Built with Streamlit, TensorFlow, and Python, the system leverages an Autoencoder deep learning model trained on large-scale transaction data to identify abnormal transaction behavior patterns.

The platform provides:

• Real-time fraud prediction
• Transaction data analysis
• Interactive visualizations
• User-friendly fraud detection interface

---

Key Features

🧠 Deep Learning Fraud Detection Engine

Uses an Autoencoder neural network to detect anomalous transaction behavior associated with fraudulent activity.

---

📊 Transaction Data Visualization

Interactive visualizations for exploring:

• Transaction distributions
• Fraud vs non-fraud patterns
• Behavioral trends in payment activity

---

⚡ Real-Time Prediction System

Allows users to input transaction details and instantly receive fraud prediction results.

---

Dataset Exploration Interface

Provides insight into transaction records, feature distributions, and sample financial data.

---

Model Overview

• Algorithm: Deep Learning Autoencoder
• Task Type: Anomaly Detection / Fraud Detection
• Framework: TensorFlow / Keras
• Input Features: Online transaction attributes
• Output: Fraudulent or Non-Fraudulent transaction prediction

The model was trained using a large-scale online payment transaction dataset containing millions of transaction records.

---

System Architecture

User Transaction Input
        ↓
Data Validation & Preprocessing
        ↓
Feature Engineering Layer
        ↓
Autoencoder Deep Learning Model
        ↓
Anomaly Detection Logic
        ↓
Fraud Prediction Output
        ↓
Streamlit Visualization Interface

##Run Locally

Clone the repo

git clone https://github.com/Pro-phet123/Final-year-work.git

Enter into project directory

cd Final-year-work

Create virtual environment

python -m venv venv

Activate virtual environment(windows)

venv\Scripts\activate

Activate virtual environment(mac/linux)

source venv/bin/activate

Install dependencies

pip install -r requirements.txt

Run the application

streamlit run main.py

🌐 Live Demo

Launch Web App

👤 Author

Olalemi Olaoluwakintan Emmanuel — Data Scientist & AI Engineer