You can deploy this model and start earning money today!

haahadragon/
Fraud-detection

Name: Fraud-detection
Availability: InStock

Views

No views yet

Fraud Detection System

A polished, production-ready Flask web application that uses a trained Random Forest model to assess transaction fraud risk from PaySim financial data.

Features

Clean fintech-style UI with dark/ivory aesthetics
Real-time fraud prediction with confidence scores
Risk level classification: High / Medium / Low / Review
Human-readable explanation of the model's decision
Full input validation with friendly error messages
Responsive design for desktop and mobile

Project Structure

fraud_app/
├── app.py                  # Flask backend
├── fraud_model.pkl         # Trained model (you provide this)
├── requirements.txt        # Python dependencies
├── README.md
├── templates/
│   └── index.html          # Main UI template
└── static/
    └── style.css           # Stylesheet

Setup Instructions

1. Clone / place the project files

Make sure all files are in the same directory (e.g. fraud_app/).

2. Place your model file

Copy your trained model into the root of the project directory:

fraud_app/fraud_model.pkl

The app expects the model at the same level as app.py.

3. Create a virtual environment (recommended)

bash

python -m venv venv
source venv/bin/activate        # Linux / macOS
venv\Scripts\activate           # Windows

4. Install dependencies

pip install -r requirements.txt

5. Run the app

python app.py

The app will start at http://127.0.0.1:5000.

Model Details

The model (fraud_model.pkl) is a Random Forest classifier trained on the PaySim synthetic financial dataset.

Feature Engineering (applied at inference time)

Feature	Formula
`balanceDiffOrig`	`oldbalanceOrg - newbalanceOrig`
`balanceDiffDest`	`newbalanceDest - oldbalanceDest`
`errorBalanceOrig`	`newbalanceOrig + amount - oldbalanceOrg`
`errorBalanceDest`	`oldbalanceDest + amount - newbalanceDest`

Transaction Type Encoding

One-hot encoded with drop_first=True. Baseline category = CASH_IN.

Expected dummy columns: type_CASH_OUT, type_DEBIT, type_PAYMENT, type_TRANSFER.

Input Fields

Field	Description
Step	Time unit in simulation (1 = 1 hr)
Transaction Type	CASH_IN, CASH_OUT, DEBIT, PAYMENT, TRANSFER
Amount	Value of the transaction
Sender Old Balance	Sender's balance before transaction
Sender New Balance	Sender's balance after transaction
Receiver Old Balance	Receiver's balance before transaction
Receiver New Balance	Receiver's balance after transaction

Risk Classification Logic

Prediction	Confidence	Risk Level
Fraud	≥ 85%	High Risk
Fraud	60–84%	Medium Risk
Fraud	< 60%	Moderate Risk
Legit	≥ 85%	Low Risk
Legit	< 85%	Review Recommended

Notes

No database or authentication required
Model is loaded once at app startup for performance
Debug mode is enabled by default; disable for production

haahadragon/Fraud-detection