Analysis of Rainfall

This project analyzes fraudulent bidding behavior (specifically, Rainfall Analysis) in online auction platforms using Exploratory Data Analysis (EDA) techniques.

📁 Project Overview

Online auction platforms are increasingly susceptible to fraudulent bidding, where users artificially drive up prices. This project uses a public dataset to:

• Compare behaviors of legitimate vs shill bidders
• Analyze bidding patterns (early, late, aggressive)
• Visualize key auction dynamics
• Propose a foundation for fraud detection using ML

📊 Dataset

• Source: Publicly available CSV titled Rainfall Dataset.csv
• Records: 6,321
• Features: 11 numeric features
• Target: Binary classification (Class - 0: Legitimate, 1: Shill)

Key Features:

• Auction_Duration
• Auction_Bids
• Winning_Ratio
• Early_Bidding
• Last_Bidding
• Class (target label)

🧪 EDA Highlights

• Class Imbalance: ~89% Legitimate, ~11% Shill Bidding
• Bidding Behavior:
  • Shill bidders place more bids on average
  • Higher early bidding concentration
  • Erratic or strategic avoidance of last-minute bids
• Winning Ratio:
  • Higher winning ratio in shorter auctions among shill bidders

📈 Tools Used

• Python
  • NumPy
  • Pandas
  • Matplotlib
  • Seaborn

🧠 Future Scope

• Machine learning classifiers (e.g., Random Forest, XGBoost)
• Real-time fraud detection pipelines
• Feature engineering from session patterns
• Behavioral profiling and anomaly detection

📚 References

• NumPy Documentation
• Pandas Documentation
• Matplotlib Documentation
• Seaborn Documentation
• LinkedIn Project Post

🧑‍🎓 Author

Kaja Revanth Sri Narasimha
Registration No: 12312200
Lovely Professional University, Department of CSE/IT

---

📅 Project Semester: January–April 2025
🎓 Course Code: INT375
👩‍🏫 Guided by: Dr. Tanima Thakur