🏡 Muzaffarpur Real Estate AI Valuation Engine (HousePredictor)

A client-side machine learning web application that predicts property valuations across key localities in Muzaffarpur, Bihar. Built with a 5-layer PyTorch Neural Network exported to ONNX format, running zero-latency inference directly inside the browser via WebAssembly (onnxruntime-web).

---

🌟 Key Features

• Zero Server Latency: Runs PyTorch model inference client-side in the browser via WebAssembly.
• Localized Feature Set: Predicts property prices based on Locality (Mithanpura, Kalyani, Motijheel, Kanhauli, Sabhapur, Sherpur, Club Road), SqFt Size, Bedrooms, Floors, and Main Road Access.
• Real-time Normalization Pipeline: Applies Z-score input feature scaling (X - mean) / std[cite: 6, 7] and un-normalizes target price predictions directly in JavaScript.
• INR Currency Formatting: Formats estimated outputs cleanly into Lakhs and Crores INR.

---

📐 Architecture Pipeline

  ┌──────────────────┐       ┌──────────────────┐       ┌──────────────────┐
  │  Form Inputs     │  ───> │  Z-Score Feature │  ───> │  ONNX Engine     │
  │  (Locality/SqFt) │       │  Normalization   │       │  (WebAssembly)   │
  └──────────────────┘       └──────────────────┘       └──────────────────┘
                                                                 │
                                                                 ▼
  ┌──────────────────┐       ┌──────────────────┐       ┌──────────────────┐
  │  Price Display   │  <─── │  INR Currency    │  <─── │  Un-normalize    │
  │  (Lakhs/Crores)  │       │  Formatter       │       │  Output Target   │
  └──────────────────┘       └──────────────────┘       └──────────────────┘