Fraud Detection System

A Java application that detects fraudulent orders by identifying multiple orders placed to the same address using different email addresses within a 24-hour time window.

Features

• Address Normalization: Intelligently normalizes addresses to detect variations of the same location

  • Handles common abbreviations (Street → St, Avenue → Ave, Road → Rd, Drive → Dr)
  • Normalizes apartment/unit/suite designations
  • Handles NYC variations (NYC → New York City)
  • Case-insensitive comparison

• Real-time Fraud Detection: Analyzes order streams to identify suspicious patterns

  • Detects different email addresses shipping to the same normalized address
  • 24-hour sliding time window for fraud detection
  • Generates alerts for suspicious orders

Requirements

• Java 8 or higher
• No external dependencies required

Project Structure

fraud-detection-system/
├── src/
│   └── Main.java          # Main application with fraud detection logic
└── README.md

How to Run

Compile

javac src/Main.java

Run

java -cp src Main

How It Works

1. Order Processing: The system receives orders containing:

   • Order ID
   • Email address
   • Shipping address
   • Timestamp

2. Address Normalization: Each address is normalized using pattern matching and string transformations to handle variations like:

   • "123 Main Street, Apt 4B, NYC"
   • "123 Main St, Apartment 4B, NYC"
   • "123 Main Street Apt 4-B New York City"

3. Fraud Detection: The system checks if:

   • Multiple orders are placed to the same normalized address
   • Orders have different email addresses
   • Orders occur within 24 hours of each other

4. Alerts: When fraud is detected, the system generates alerts linking suspicious orders.

Example Output

FRAUD ALERTS:
FRAUD ALERT: ORD003 - Matches previous order ORD001 (different email, same address)
FRAUD ALERT: ORD005 - Matches previous order ORD001 (different email, same address)

Configuration

You can modify the time window for fraud detection by changing the TIME_WINDOW_HOURS constant in the Main class (default: 24 hours).

Code Structure

Classes

• Order: Represents an incoming order with ID, email, address, and timestamp
• OrderInfo: Lightweight order information stored for comparison
• Main: Contains the fraud detection logic and address normalization

Key Methods

• normalizeAddress(String address): Normalizes address strings for comparison
• detectFraud(List<Order> orders): Analyzes orders and returns fraud alerts
• processOrdersInRealTime(List<Order> stream): Processes order stream and prints alerts

License

This project is for educational purposes.