💳 Credit Card Fraud Detection App

A Streamlit web application that uses a trained XGBoost classification model to detect the risk of credit card fraud based on anonymized transaction features.

---

🚀 Live App

👉 Click here to access the live Streamlit app

---

📊 About the Dataset

• Source: Kaggle — Credit Card Fraud Detection

• Samples: 284,807 transactions

• Fraud Cases: 492 (~0.17%) — heavily imbalanced

• Features Used (Top 5 selected based on importance):

  • V14: PCA component strongly correlated with fraud
  • V10: Linked to unusual transaction behavior
  • V12: Reflects customer spending behavior
  • V4: Captures transaction frequency signals
  • V17: Often tied to internal risk scores

• Target Variable: Class

  • 0: Genuine Transaction
  • 1: Fraudulent Transaction

---

🛠️ How It Works

Users input 5 anonymized PCA-derived features from a credit card transaction. The app:

• Loads a pre-trained XGBoost model (fraud_xgb_model.pkl)
• Predicts whether the transaction is fraudulent or genuine
• Displays the fraud probability score

---

🎯 App Features

• ✅ Clean Streamlit UI with input tooltips and examples
• ✅ Real-time fraud prediction with risk score
• ✅ Educates users on dataset and model behavior
• ✅ Clear disclaimers about limitations

---

⚠️ Model Disclaimer

• This model is trained on anonymized and imbalanced data.

• While tuned with SMOTE and hyperparameter optimization, misclassifications may still occur.

• Designed for:

  • Educational use ✅
  • Demonstrating end-to-end ML ✅
  • Prototyping and portfolios ✅

• ❌ Not suitable for production or real-time banking environments.

---

🧪 Example Inputs

Feature	Placeholder Example
V14	-2.3456
V10	3.2178
V12	-4.1023
V4	1.2567
V17	-1.7890

---

📝 Technologies Used

• Python
• Streamlit
• XGBoost
• scikit-learn
• pandas, numpy
• joblib

---

📁 Project Structure

Fraud_Detection/
├── app.py                  # Streamlit frontend app
├── fraud_xgb_model.pkl     # Trained ML model
├── requirements.txt        # Dependencies
└── README.md               # Project documentation

---

📦 Requirements

Install dependencies using:

pip install -r requirements.txt