⚡ GridWatch — AI-Driven Energy Theft & Anomaly Detection System

24-Hour Hackathon | Energy Conservation Week 2026

---

🚀 Setup & Run (Windows)

Step 1 — Make sure Python is installed

Open Command Prompt and run:

python --version

You need Python 3.9 or above. Download from https://python.org if needed.

---

Step 2 — Install all dependencies

Open Command Prompt INSIDE the gridwatch folder, then run:

pip install -r requirements.txt

This installs everything. Takes 1-2 minutes.

---

Step 3 — Launch the dashboard

streamlit run dashboard.py

Your browser will open automatically at http://localhost:8501

---

🎮 How to Demo It

1. The dashboard loads with 30 days of pre-generated historical data
2. Click "Simulate New Readings" in the sidebar to push live data
3. Watch the Live Dashboard — anomalies appear in red within seconds
4. Go to Consumer Drilldown → select C003, C007, or C009 to see theft patterns
5. Go to Alert History to see all flagged cases with risk scores

The 3 "Theft" Consumers (for demo):

Consumer	ID	Anomaly Type
Mohammed Iqbal	C003	Meter tampered — usage drops to 20% of normal
Arjun Patel	C007	Meter bypassed — near-zero readings
Venkat Raman	C009	Gradual decline — slow tamper over time

---

📁 File Structure

gridwatch/
├── dashboard.py        ← Main Streamlit app (run this)
├── model.py            ← Isolation Forest ML model
├── data_generator.py   ← Synthetic smart meter data
├── database.py         ← SQLite database layer
├── requirements.txt    ← All dependencies
└── README.md           ← This file

---

❓ Troubleshooting

"streamlit is not recognised" → Run: pip install streamlit then try again

Dashboard is slow on first load → Normal — it's generating 30 days of historical data and training ML models. Takes ~15 seconds once, then it's fast.

Port already in use → Run: streamlit run dashboard.py --server.port 8502

---

Built with Python, Scikit-learn, Streamlit, SQLite, Plotly — 100% free & open source