Credit Card Fraud Detection

🚀 Overview

This project aims to detect fraudulent credit card transactions using machine learning techniques. Given the highly imbalanced nature of fraud detection, we apply various resampling techniques and model evaluation metrics to ensure robust predictions.

📂 Dataset

The dataset used for this project is the Credit Card Fraud Detection dataset available on Kaggle. It consists of:

• 284,807 transactions
• 492 fraudulent transactions (0.172% of total data)
• Features: 30 columns (V1-V28 are PCA-transformed), Time, Amount, and Class (target variable: 0 = Legitimate, 1 = Fraudulent)

🛠️ Approach

1️⃣ Data Preprocessing

• Handling class imbalance using SMOTE (Synthetic Minority Over-sampling Technique) and undersampling.
• Scaling numerical features using StandardScaler or MinMaxScaler.
• Feature engineering to extract meaningful transaction patterns.

2️⃣ Exploratory Data Analysis (EDA)

• Distribution of transaction amounts and time.
• Fraudulent vs. legitimate transaction patterns.
• Correlation analysis of PCA-transformed features.

3️⃣ Model Selection & Training

We experimented with various machine learning models: ✅ Random Forest ✅ AdaBoost ✅ XGBoost ✅ Lightbgm ✅ Neural Networks (Deep Learning Approach)

4️⃣ Evaluation Metrics

Since fraud detection is an imbalanced classification problem, we focus on:

• Precision, Recall, F1-score (to minimize false negatives)
• AUC-ROC Curve (to evaluate the model’s discriminatory power)
• Confusion Matrix (to analyze misclassification rates)

☁️ Deployment with AWS SageMaker

This project utilizes AWS SageMaker for deploying the trained fraud detection model. The deployment steps include:

1️⃣ Model Training on SageMaker

• Using built-in SageMaker algorithms or custom scripts.
• Training the model with SageMaker’s managed Jupyter notebooks.

2️⃣ Model Deployment

• Deploying the trained model as a real-time endpoint.
• Using SageMaker Inference for making predictions on new transactions.

📊 Results

• The best model achieved:
  • 88.1% roc auc score
  • 85.2% F1-score
• Random Forest and XGBoost were the most balanced models in terms of performance.
• Neural Networks performed exceptionally well in high-computational environments with AUC close to 0.999.

🔮 Future Improvements

• Implementing deep learning architectures for better accuracy.
• Deploying the model using Flask or FastAPI for API-based fraud detection.
• Enhancing real-time fraud detection using streaming data (e.g., Apache Kafka).

🤝 Contributing

Contributions are welcome! Feel free to submit pull requests or raise issues.

📝 License

This project is licensed under the MIT License.

📢 Acknowledgments

• Open-source Machine Learning & Data Science Communities

---

⭐ Feel free to star this repository if you find it useful! 🌟